Skip to content
dialegein

Corrections

Corrections

This index is built on the claim that its evidence can be checked. A site that records what it read and then never mentions a mistake would be using that evidence as decoration, so here is the other half: every statement we published that turned out to be wrong, what it says now, and how it was caught.

There are 546 of them so far, and the list grows as the rest of the programme is written up. Every entry is a statement a reader could have met, and internal housekeeping, refactors and tooling fixes are deliberately absent, because if a reader never saw the wrong version then it is not a correction to anything. Every entry names the commit that fixed it, and that is for our own accounting rather than yours: the repository is not public, so you cannot open the commit. This page therefore asks you to take one thing on trust that the rest of the site does not, and it is worth saying so plainly rather than implying otherwise. What you can check is the result, because the records, the pages and the dataset are all live: a vendor either says the thing we quoted or does not.

The list

  1. 15ea638

    The sitemap index freshness, /status/ lastmod churn, hub metadata, mobile and comparison handoffs, product job links, /tools/ H1, finder no-JS fallback, shared social-card alt and small touch targets

    What it said
    The sitemap index lastmod was 2026-10-18 while the child sitemap newest was 2026-10-19; /status/ rendered a relative age that changed on every build, breaking the page-lastmod fixpoint; six hub descriptions stopped mid-phrase; product job chips linked to canonicalised /tools/?job= URLs instead of the job hubs; comparison pages had no direct vendor handoff; the mobile product page put its first Visit action 75-84% down the page; the Finder no-JS fallback exposed zero rows; /tools/ shared the home H1; the shared social-card alt described the homepage on every route; and the clear-search and directory controls were under the 24px touch target.
    What it says now
    sitemap-index.xml is synced to the child shard and check-sitemap-dates rejects an older index; /status/ age is normalised for hashing; the six descriptions end cleanly; product job chips use /jobs/<slug>/; comparison record blocks include Visit through /go/<slug>/ and conditional alternatives links; mobile product pages show Visit near the top; the Finder no-JS fallback shows all server-rendered results; /tools/ says All PDF tools; the shared social alt describes the shared image; and the small controls have a 24px minimum target. The evidence-backed backlog is docs/seo-editorial-backlog-2026-10-19.md. No third-party tracker, speculative page or external change was added.

  2. 99b54b5

    The ICEpdf websiteUrl, source metadata, watchOutFor URL/source-age wording, claim source URLs/checkedAt and the old platform-candidates review note

    What it said
    websiteUrl used the non-canonical https://icepdf.org/, which redirects through http://www.icepdf.org/ to https://www.icepdf.org/. source metadata still described a 2026-09-22 read of icepdf.org rather than the canonical live page and current access. The watchOutFor did not clearly separate the live URL from stale official product documentation, and the old platform-candidates review still recorded the URL as an unresolved item with the final page unfetched.
    What it says now
    websiteUrl is https://www.icepdf.org/ (final 200; canonical host). source metadata records the 2026-10-18 redirect chain, current access, the 4.2.2 release notes dated 23-08-2011, the 22-08-2011 Developer Guide, the 2009 footer and the absence of any official code-host link. watchOutFor says the URL is live but the official documentation is stale, does not call the project abandoned, and qualifies the independently identified pcorless GitHub repository as not an official-site source link. Claims retain their current wording, use https://www.icepdf.org/ and carry checkedAt 2026-10-18. The old review note is marked resolved with the same evidence. lastReviewedAt remains null because no human review is claimed; owner/licence/version/platform/jobs were left unchanged because the URL audit did not establish different facts.

  3. 11a0b2f

    The pdf-xchange-sdk, libreoffice-draw and pdfelement websiteUrl/source metadata

    What it said
    Three cited canonical URLs no longer served their recorded product-specific pages: tracker-software.com redirected to the pdf-xchange.com product page, libreoffice.org/discover/draw/ redirected to the site home/screenshots, and pdf.wondershare.com/pdfelement/ redirected to the Wondershare home. The records still used the old destinations.
    What it says now
    websiteUrl is now https://www.pdf-xchange.com/product/pdf-xchange-core-api-sdk for pdf-xchange-sdk, https://www.libreoffice.org/ for libreoffice-draw (the dedicated Discover/Draw page is gone), and https://pdf.wondershare.com/ for pdfelement (the dedicated product path canonicalizes to the brand home). Source metadata records the 2026-10-18 redirect evidence, and no owner/licence/version/platform/job claim was re-derived from the redirect. PandaDoc 429 and Spire 403 were treated as access/rate-limit responses, not broken URLs.

  4. 87061b5

    The sitemap date source and guard, product title/meta-description overrides, the optional description field, the all-product snippet audit and the public/_redirects /go/ comments

    What it said
    The sitemap-index advertised 2027-11-11 because recordLastmod scanned every ISO date in a record, including a cookie-expiry date in jpgtopdf internalNotes. Seventeen generated product titles ended mid-phrase, including reportlab (\"for generating\"), gimagereader (\"that runs\") and pdfrest (\"built\"); bluebeam-revu and telerik-dpl descriptions were cut mid-clause. The public/_redirects comments still said the build had no /go/ click endpoint and left it to 404, although the Worker serves /go/<slug> with a 302 and noindex/nofollow.
    What it says now
    recordLastmod uses only explicit evidence dates (lastReviewedAt, claims checkedAt, edition readAt, and a source date only after read/accessed/reviewed/checked), refuses dates later than the UTC build day and ignores internalNotes. sitemap-index.xml and sitemap-0.xml both have newest lastmod 2026-10-18, and a new build/verify guard rejects future or malformed sitemap dates. The slashed child loc is kept because /sitemap-0.xml/ is a verified 200 rewrite and patch-sitemap-index checks it. Seventeen product records now carry complete titles and ten priority records plus telerik-dpl carry fact-supported descriptions; the schema and product template support a nullable description override. A new all-155-product snippet audit rejects empty, overlong and stop-word/dangling-ending snippets. The /go/ comments now describe the 302 handoff, noindex/nofollow and robots disallow. No /go/, query-canonical or redirect behavior changed.

  5. 11a5a31

    The 17 product title overrides added in the SEO snippet repair

    What it said
    Every newly added title override contained the Unicode replacement character U+FFFD between the word review and the evidence-backed title fragment, so the rendered title showed a replacement glyph instead of a separator.
    What it says now
    Each of the 17 title overrides uses a plain ASCII hyphen separator, written as review - fragment, with the evidence-backed wording unchanged: ABBYY FineReader PDF, axesCheck, axesPDF, Bluebeam Revu, Adobe PDF Library, DocRaptor, gImageReader, LibreOffice Draw, Marker, Microsoft Edge PDF reader, pdfminer.six, pdfRest, PyMuPDF, ReportLab, SimplePDF, tc-lib-pdf and WeasyPrint. A diff against 6abe9bc has no newly added line containing U+FFFD.

  6. 7dbf343

    The Microsoft Edge PDF reader profile, in the public watchOutFor field

    What it said
    The sentence about digital signatures carried the Unicode replacement character U+FFFD between signatures and only view and validate them, so the page rendered a replacement glyph instead of punctuation.
    What it says now
    The sentence uses a plain ASCII hyphen separator: it cannot create digital signatures - only view and validate them. The wording and the recorded evidence are unchanged, and a scan of all 155 product records finds no remaining literal U+FFFD. Historical audit notes and documentation outside product records were left untouched.

  7. 7a053ac

    The Genius Scan profile, in the platforms and interfaces fields

    What it said
    It listed a platform of api and an interface of library, which described the SDK that The Grizzly Labs licenses separately, not the scanner app the profile is about.
    What it says now
    Platforms are web, ios and android, and interfaces is empty. The app is a mobile scanner with a web app, and the SDK is a separate offering on the same site.

  8. 7a053ac

    The AvePDF tagline

    What it said
    It read Free EU browser PDF toolkit. EU in a short product label can be read as where documents are processed, while the dataResidency field says eu and the processesLocallyNote says that needs a direct regional source, so the geography did no useful work and could imply more than the data supports.
    What it says now
    Free browser PDF toolkit, operated by the company behind the Nutrient SDK. Operator identity and capability unchanged, and no residency claim replaced the token.

  9. 7a053ac

    The AvePDF record, in the data residency field

    What it said
    The field said eu, while the record own evidence note says the documented server upload and processing is what the record can evidence and that EU residency would still need a direct regional source. An operator address in France and a hosting entry naming a provider do not establish where uploaded documents are processed or stored, and permission for international transfers is not evidence of multi region processing.
    What it says now
    unknown, which the schema already supports. No region was invented, operator jurisdiction was not conflated with data residency, and every retention and model training caveat is unchanged.

  10. 7a053ac

    The ReportLab record, in the primary convert job

    What it said
    The primary convert job carried no scope, so a reader saw the same job label used for Word, Excel and image conversion on a library whose vendor page describes generating bespoke PDFs from code, with a hosted platform alongside, and never describes converting a Word, Excel or image file.
    What it says now
    The job keeps convert and gains the scope PDF generation from code; not a Word, Excel or image converter, which is what https://www.reportlab.com/ supports, read on 2026-10-16. The tagline, prose, pricing and licence fields are unchanged.

  11. 7a053ac

    The borb record, in the primary convert job

    What it said
    The primary convert job carried no scope, so the job label used elsewhere for Word, Excel and image conversion sat on a Python library whose own page headlines PDF creation and offers conversion only through named add-ons such as borb_markdown_to_pdf and borb_ocr.
    What it says now
    The job keeps convert and gains the scope PDF creation plus Markdown-to-PDF and OCR add-ons; not a Word or Excel file converter, which is the boundary https://borbpdf.com/ draws, read on 2026-10-16. Tagline, prose, licence and pricing fields are unchanged.

  12. 7a053ac

    The PDFShift record, in the primary convert job

    What it said
    The primary convert job carried no scope, so a job label used for Word, Excel and image conversion sat on a hosted API whose own title describes converting websites and HTML to PDF.
    What it says now
    The job keeps convert and gains the scope HTML and website capture to PDF through an API; not a Word, Excel or image converter, which is what https://pdfshift.io/ states on 2026-10-16. Tagline, prose, pricing and licence fields are unchanged.

  13. 7a053ac

    The mPDF record, in the primary convert job

    What it said
    The primary convert job carried no scope, so a job label used for Word, Excel and image conversion sat on a PHP library whose own manual says it generates PDF files from UTF-8 encoded HTML and is not meant to receive HTML or CSS from an outside user.
    What it says now
    The job keeps convert and gains the scope UTF-8 HTML and CSS to PDF; not a Word, Excel or image converter, which is what https://mpdf.github.io/ states on 2026-10-16. The tagline, prose, licence and pricing fields are unchanged, including the record own note that the manual calls the library dated.

  14. 7a053ac

    The FPDI record, in the primary convert job

    What it said
    The primary convert job carried no scope, so a label used for Word, Excel and image conversion sat on a PHP class library whose vendor page describes importing pages from existing PDFs and using them as templates.
    What it says now
    The job keeps convert and gains the scope importing pages from existing PDFs into a template; not a Word, Excel or image converter, which is what https://www.setasign.com/products/fpdi/ states on 2026-10-16. Tagline, licence and pricing fields are unchanged, and no capability was added.

  15. 7a053ac

    The Adobe Acrobat record, in the primary convert job

    What it said
    The primary convert job carried no scope, so a reader could not tell which direction and which inputs the conversion covers, on a record whose page divides creating PDFs from exporting them.
    What it says now
    The job keeps convert and gains the scope Word, PowerPoint, Excel, Publisher, text, image and other supported files to PDF in Reader; export is a separate feature, which is what https://helpx.adobe.com/reader/desktop/create-pdf.html documents on 2026-10-16. The tagline and every other field are unchanged, and the merge job was not touched.

  16. 7a053ac

    The Apache PDFBox record, in the secondary edit job

    What it said
    The edit job carried no scope, so a reader had no statement of what manipulating means here, on a record whose project page separates creation, manipulation of existing documents and content extraction.
    What it says now
    The edit job keeps its mapping and gains the scope manipulating existing PDF documents in the Java library, which is what https://pdfbox.apache.org/ states on 2026-10-16. The scope does not mention redaction in either direction, because the project feature list does not offer it and the record already records that it is unavailable. Convert, merge, compress and all other fields are unchanged.

  17. 7a053ac

    The pypdf record, in the secondary edit job

    What it said
    The edit job carried no scope, so a reader had no statement of which page operations it covers, on a library whose documentation separates rotation, view box cropping and content transformation, and warns that cropping does not remove what it hides.
    What it says now
    The edit job keeps its mapping and gains the scope rotating and cropping pages by view box, and transforming or merging page contents; cropping hides content rather than removing it, which is what https://pypdf.readthedocs.io/en/latest/user/cropping-and-transforming.html documents on 2026-10-16. The scope mentions no redaction and no irreversible deletion, because the source offers neither and warns against the assumption.

  18. 7a053ac

    The pdf-lib record, in the secondary merge job

    What it said
    The merge job carried no scope, so a reader had no statement of what merging covers on a library whose project page separates split and merge from creation and modification.
    What it says now
    The merge job keeps its mapping and gains the scope merging and splitting PDF pages in JavaScript, which is what https://pdf-lib.js.org/ states on 2026-10-16. The record own encrypted input limitation is unchanged and the scope implies no decryption.

  19. 7a053ac

    The pdfcpu record, in the secondary edit job

    What it said
    The edit job carried no scope, so a reader had no statement of which editing operations the tool supports, on a Go library whose package documentation enumerates its commands.
    What it says now
    The edit job keeps its mapping and gains the scope page insertion, removal and rearrangement, stamps, watermarks, forms, bookmarks and document properties, taken from the command list at https://github.com/pdfcpu/pdfcpu/blob/master/pkg/pdfcpu/doc.go on 2026-10-16. That list contains no redaction and no free text editing command, so the scope names neither. Merge, compress, convert, protect and all other fields are unchanged.

  20. 7a053ac

    The qpdf record, in the merge job

    What it said
    The merge job carried no scope, so a reader had no statement of what merging covers on a tool whose manual separates structural transformation from content modification.
    What it says now
    The job keeps merge and gains the scope structural merging and splitting of PDF pages, and collating, from the Running qpdf chapter at https://qpdf.readthedocs.io/en/latest/cli.html read on 2026-10-16. The word structural is the manual own distinction, and the scope implies no page content editing. Compress, protect and platforms are unchanged.

  21. 7a053ac

    The Poppler record, in the primary convert job

    What it said
    The convert job carried no scope, so a reader had no statement of which conversions the toolset covers, on a rendering library whose project page names frontends rather than output formats.
    What it says now
    The job keeps convert and gains the scope rendering and converting PDF pages to image, HTML and PostScript output through the bundled utilities, from https://poppler.freedesktop.org/ read on 2026-10-16. It implies neither Office file conversion nor a hosted web application. Extract, view, merge and platforms are unchanged.

  22. 7a053ac

    The pdftk-java record, in the merge job

    What it said
    The merge job carried no scope, so a reader had no statement of what merging covers on a tool whose own description separates merging and collating from splitting and from its other operations.
    What it says now
    The job keeps merge and gains the scope merging and collating PDF documents, and splitting pages, taken from the tool description at https://gitlab.com/pdftk-java/pdftk/-/raw/master/java/com/gitlab/pdftk_java/resources/description.txt read on 2026-10-16. It implies no page content editing. Protect, edit and platforms are unchanged.

  23. 7a053ac

    The Apache PDFBox record, in the secondary convert job

    What it said
    The convert job carried no scope, so a reader had no statement of which direction of conversion the tool covers, on a library whose feature list names creating PDFs and saving PDFs as images rather than converting other formats.
    What it says now
    The job keeps convert and gains the scope creating PDFs from scratch with embedded fonts and images, and rendering PDFs to images such as PNG or JPEG, taken from https://pdfbox.apache.org/ read on 2026-10-16. It implies neither Office file conversion nor a hosted browser converter. Edit, merge, compress, protect and all other fields are unchanged.

  24. 7a053ac

    The pdfcpu record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader had no statement of what compression means here, on a tool whose command reference describes optimize as removing redundant page resources rather than as a size outcome.
    What it says now
    The job keeps compress and gains the scope optimizing PDFs by removing redundant page resources, taken from https://raw.githubusercontent.com/pdfcpu/pdfcpu/master/pkg/pdfcpu/doc.go read on 2026-10-16. It claims no lossless guarantee, no target ratio and no image recompression, because the source states none. Edit, merge, convert, protect and platforms are unchanged.

  25. 7a053ac

    The pikepdf record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader had no statement of what compression means here, on a library whose API documents specific optimisation and stream controls rather than a size outcome.
    What it says now
    The job keeps compress and gains the scope optimising PDFs by removing unreferenced resources, recompressing streams and optimising images, taken from the documented API surface at https://pikepdf.readthedocs.io/en/stable/api/main.html read on 2026-10-16. It promises no size reduction and no lossless result. Merge, platforms and all other fields are unchanged.

  26. 7a053ac

    The PyMuPDF record, in the secondary convert job

    What it said
    The convert job carried no scope, so a reader could not tell whether the conversion covers the core package, the separate PyMuPDF Pro product or the external libraries that also use it.
    What it says now
    The job keeps convert and gains the scope converting supported document types such as Markdown and XPS to PDF, and rendering PDF pages to images and SVG, all from https://pymupdf.readthedocs.io/en/latest/converting-files.html read on 2026-10-16. The page attributes PDF to DOCX to the separate pdf2docx library and PDF to Markdown to PyMuPDF4LLM, and lists PyMuPDF Pro as a separate product, so the scope stays within the core package and implies no Office conversion there and no hosted service. Merge, compress and platforms are unchanged.

  27. 7a053ac

    The MuPDF mutool record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader had no statement of which compression controls the tool offers.
    What it says now
    The job keeps compress and gains the scope rewriting PDFs with image subsampling and recompression controls, including lossless, jpeg, j2k, fax and jbig2 methods, taken from https://mupdf.readthedocs.io/en/1.27.2/tools/mutool-clean.html read on 2026-10-16. It promises no size reduction. The scope deliberately omits garbage collection and duplicate-object reuse, because the fetch truncated the option list where those are documented and I did not read them. Convert, merge, edit, protect, sign and platforms are unchanged.

  28. 7a053ac

    The Adobe Acrobat record, in the secondary merge job

    What it said
    The merge job carried no scope, so a reader had no statement of what combining covers or where it happens.
    What it says now
    The job keeps merge and gains the scope combining files into one PDF in the desktop application, taken from https://helpx.adobe.com/in/acrobat/desktop/edit-documents/combine-files/combine-files.html read on 2026-10-16, whose title reads Combine files into one PDF in Acrobat and which sits under the Acrobat desktop help. The fetch truncated the body, so the scope names no input formats, because the supported types list was not read. Convert is unchanged and edit has no scope by design.

  29. 7a053ac

    The Antenna House Formatter record, in the convert job

    What it said
    The convert job carried no scope, so a reader had no statement of which inputs and styling languages the conversion covers.
    What it says now
    The job keeps convert and gains the scope XML, DITA and HTML styled with XSL-FO or CSS to PDF, taken from https://www.antennahouse.com/formatter-v7 read on 2026-10-16, whose title is Convert XML or HTML to PDF and which states publishing high quality PDF from XML, DITA and HTML with content styled by XSL-FO or CSS. It does not broaden to Office conversion and claims no hosted service.

  30. 7a053ac

    The Apache Tika record, in the secondary ocr job

    What it said
    The ocr job carried no scope, so a reader could not tell whether Tika performs OCR itself or hands documents to an external engine.
    What it says now
    The job keeps ocr and gains the scope optional OCR for images and PDF pages when an external OCR engine such as Tesseract is installed and configured, taken from https://tika.apache.org/docs/4.0.x/formats.html read on 2026-10-16. That page says PDFParser can hand pages or images to OCR or a vision model, that text inside images is not extracted by the standard image parsers, and that external tools such as ffmpeg are available when installed. The fetch truncated the OCR section body, so the scope says an engine such as Tesseract rather than asserting a Tesseract-specific integration, and claims no built-in OCR and no hosted processing.

  31. 7a053ac

    The Adobe PDF Extract API record, in the secondary convert job

    What it said
    The convert job carried no scope, so a job label used for converting between file formats sat on a cloud extraction service whose outputs are renditions of the extracted content rather than converted documents.
    What it says now
    The job keeps convert and gains the scope renditions only: structured JSON or Markdown output, with tables optionally as CSV or XLSX and images as PNG; not general PDF conversion, taken from https://developer.adobe.com/document-services/docs/overview/pdf-extract-api/ read on 2026-10-16, whose wording is that table data can optionally be delivered in CSV and XLSX files and images are extracted as PNG files. Extract stays primary and unscoped, and ocr is unchanged.

  32. 7a053ac

    The Adobe Scan record, in the local processing wording versus attribute signal

    What it said
    The sweep flagged the record prose about exporting or copying a scan to the device as sitting against attributes.processesLocally being false, which reads like a contradiction.
    What it says now
    Nothing in the record changed, because there is no contradiction to fix. Adobe own Android documentation says, under Copy files to device, that when you save a scan it automatically uploads to Document Cloud, and only then offers Copy to device to save a copy locally; it notes that local PDFs appear with an offline badge; and under Save files to the cloud it repeats that a saved scan uploads to Document Cloud automatically before third party clouds are mentioned. Sharing is also described as sharing a link to a file stored in Adobe Document Cloud. So copying to the device is a destination for a finished file, while capture, cleanup, recognition and default storage remain cloud backed, which is exactly what processesLocally false states. The disposition is verified-current with no content change.

  33. 7a053ac

    The Apache PDFBox record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader had no statement of what compression means for this library.
    What it says now
    The job keeps compress and gains the scope compressed saving of documents by default, overridable per save; not a standalone PDF optimiser, from https://pdfbox.apache.org/3.0/migration.html read on 2026-10-16, which says under Compressed saving by default that saving a PDF is now done in compressed mode by default and that PDDocument.save with CompressParameters.NO_COMPRESSION overrides it. The scope promises no size outcome, and deliberately says nothing about render time image options, which compress images in the output being rendered rather than rewriting the input PDF. Edit, merge, convert and protect are unchanged.

  34. 7a053ac

    The Apple Preview record, in the secondary convert job

    What it said
    The convert job carried no scope, so a reader had no statement of what conversion means on an app whose documentation describes exporting a file as another type and converting between image formats.
    What it says now
    The job keeps convert and gains the scope exporting a PDF or image as another file type, and converting between image formats such as HEIC, JPEG, JPEG-2000, PNG and TIFF, in the Mac app, from https://support.apple.com/en-gb/guide/preview/prvw0e8da223/mac read on 2026-10-16, which says you can export a PDF or image to save it as a different type of file. Both pages are Mac guides, so no iPhone or iPad parity is claimed, and nothing implies arbitrary document conversion. View stays primary.

  35. 7a053ac

    The Apple Preview record, in the secondary protect job

    What it said
    The protect job carried no scope, so a reader had no statement of what protection means in an app whose guide describes passwords and permissions rather than encryption detail.
    What it says now
    The job keeps protect and gains the scope password protection and permissions for opening, printing, copying and annotating, in the Mac app, from https://support.apple.com/en-mk/guide/preview/prvw587dd90f/mac read on 2026-10-16. That guide documents a password required to open the PDF, an owner password and permissions for printing, copying text and adding annotations, set through File then Export, Permissions, Apply and Save, taking effect the next time the file is opened. The scope claims no encryption strength, no redaction and no iPhone or iPad parity. View stays primary and convert keeps its scope.

  36. 7a053ac

    The Apple Preview record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader had no statement of what compressing a PDF means in Preview or what it may cost in quality.
    What it says now
    The job keeps compress and gains the scope reducing PDF file size through export options and the Reduce File Size Quartz filter, with possible quality loss and variable results, in the Mac app, from https://support.apple.com/en-au/guide/preview/prvw1509/mac read on 2026-10-16. That guide says that to save storage space you can compress a PDF in Preview, that a compressed PDF may be of lower quality, and that the resulting size and image quality depend on the options chosen, documenting File, Export, the Reduce File Size Quartz filter, linearized PDF creation, optimizing images for screen, and saving images as JPEG. No amount of reduction is promised, compression is not called lossless, and no iPhone or iPad parity is claimed. View stays primary and other jobs are unchanged.

  37. 7a053ac

    The Apryse record, in the secondary convert job

    What it said
    The convert job carried no scope, so a reader could not tell which conversions the documented server and desktop product actually offers, or whether every deployment shape offers them.
    What it says now
    The job keeps convert and gains the scope document conversion between PDF, XPS, EMF and WMF, SVG and raster images, including HTML to PDF, XPS or SVG, through the server and desktop SDK and its conversion tools, from https://docs.apryse.com/core/conversion/conversion and https://docs.apryse.com/documentation/samples/node/js/ConvertTest/ read on 2026-10-16. The first is titled Document conversion library for Server/Desktop and lists direct conversions and CLI tools including PDF2Image, XPSConvert, PDF2XPS and PDF2SVG; the second describes conversion between PDF, XPS, SVG, TIFF, PNG, JPEG and other image formats with built in MS Office conversion. The scope claims no WebViewer or mobile parity, no arbitrary document conversion and no free conversion service. Edit stays primary.

  38. 7a053ac

    The Apryse record, in the secondary merge job

    What it said
    The merge job carried no scope, so a reader could not tell what merging covers or which deployment shape the documentation describes it for.
    What it says now
    The job keeps merge and gains the scope merging or inserting PDF pages, through the server and desktop SDK, from https://docs.apryse.com/core/page-manipulation/merge read on 2026-10-16, whose title reads How to merge or insert PDF pages on Server/Desktop under the Server SDK. The page body was truncated, so PDFDoc.InsertPages and page range detail are deliberately not asserted, and a direct request to the page manipulation guide returned 403 Forbidden. The scope claims no page ranges, no arbitrary file joining, no Office merging and no parity with the WebViewer manipulation demo, which is a separate surface and was not used as evidence. Edit stays primary.

  39. 7a053ac

    The Apryse record, in the secondary ocr job

    What it said
    The ocr job carried no scope, so a reader could not tell whether OCR is part of the SDK, an add-on, or available across every deployment shape.
    What it says now
    The job keeps ocr and gains the scope OCR through the server and desktop SDK and its optional OCR module, producing searchable text from images or PDFs, from https://docs.apryse.com/core/ocr/ocr read on 2026-10-16, whose title reads OCR Library for Server/Desktop under the Server SDK. The body was truncated, so the three OCR options, the downloadable module across Windows, Linux and macOS, and the searchable PDF versus JSON and XML position data detail are attributed to Calum check of the same page rather than to wording read here. The scope claims no WebViewer or mobile parity and no built-in availability in every package, and keeps OCR output distinct from generic text extraction. Edit stays primary.

  40. 7a053ac

    The Apryse record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader could not tell what optimisation the Server/Desktop product offers or whether the web surface shares it.
    What it says now
    The job keeps compress and gains the scope optimising PDF file size through the server and desktop PDF Optimizer: redundant resource removal, data stream and image compression, and optional downsampling, from https://docs.apryse.com/core/optimization/optimization read on 2026-10-16, whose title reads Optimize PDF file size on Server/Desktop under the Server SDK. The body was truncated, so the operation list and the note that the Web SDK has separate save and optimization documentation are attributed to Calum check of the same page rather than to wording read here. The scope promises no lossless output and no fixed size reduction, and claims no universal WebViewer or mobile parity. Edit stays primary.

  41. 7a053ac

    The Aspose.PDF record, in the secondary merge job, and the scope check that read the rendered page

    What it said
    The merge job carried no scope, so a reader could not tell which edition the documented merging applies to. The scope check then reported aspose-pdf:merge as unrendered, because it compared the built page with a raw substring test and Astro escapes an apostrophe as &#39;.
    What it says now
    The job keeps merge and gains the scope merging or concatenating PDF documents by adding one document pages to another, in Aspose.PDF for .NET, from https://docs.aspose.com/pdf/net/merge-pdf-documents/ read on 2026-10-16, whose page says it shows how to merge multiple PDF files into a single PDF document using Aspose.PDF for .NET and that the first file is joined at the end of the other, with the sample calling document1.Pages.Add(document2.Pages) then Save. check-job-scopes.mjs now decodes the entity set Astro emits before comparing rendered text, so a genuine rendering omission still fails while an escaped apostrophe no longer does. The scope names .NET and claims no cross-edition parity, no arbitrary file joining and no Office merging; edit stays primary.

  42. 7a053ac

    The Aspose.PDF record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader could not tell what optimisation means in this library or whether a smaller file is promised.
    What it says now
    The job keeps compress and gains the scope PDF size optimisation and compression through Aspose.PDF for .NET Optimize and OptimizeResources APIs, including resource cleanup and optional image or stream compression; size reduction is not guaranteed, from https://docs.aspose.com/pdf/net/optimize-pdf/ read on 2026-10-16. That page documents Optimize for linearization for the web, OptimizeResources removing unused resources, joining equal resources and deleting unused objects, and says plainly that the method cannot guarantee document shrinking; it also documents image quality and resolution options, unused object and stream removal, duplicate stream linking, page content reuse, font unembedding and FlateDecode content stream compression. The not guaranteed clause is the vendor own wording. The scope names .NET and claims no hosted service and no lossless output. Edit stays primary and merge keeps its scope.

  43. 7a053ac

    The Aspose.PDF record, in the view job

    What it said
    The record listed a view job for Aspose.PDF, which asserts the library can be used to read or view PDF files. The vendor documentation contradicts that for viewing: it calls Aspose.PDF for .NET a PDF document creation engine and says that to view PDF documents produced with it the machine needs at least an Adobe Acrobat Reader.
    What it says now
    The view job is removed. Nothing replaced it, so the record no longer claims a viewer it does not have, and no viewer capability was implied instead. The primary edit job is unchanged, and the merge and compress scopes added on 2026-10-16 are unchanged. Evidence is https://docs.aspose.com/pdf/net/system-requirements/ read on 2026-10-16; the merge and compress pages read the same day document creation, manipulation, optimisation and conversion APIs rather than a reader. Because the evidence is .NET system requirements, a viewer evidenced for another edition would justify reinstating the job.

  44. 7a053ac

    The AvePDF record, in the convert job

    What it said
    The convert job carried no scope, so a reader could not tell which formats and directions the browser converter covers, or that the limits differ between tools.
    What it says now
    The job keeps convert and gains the scope browser-based conversion of more than 100 file types to PDF, plus documented PDF-to-image conversion tools; limits and supported formats vary by tool, from https://avepdf.com/convert-to-pdf and https://avepdf.com/pdf-to-jpeg read on 2026-10-16. Those pages say the browser converter accepts more than 100 formats and outputs PDF with a 128 MB maximum on that tool, list image, office and text, CAD and SVG inputs, note that files are uploaded for processing, and document PDF to JPG with the same 128 MB maximum. The scope claims no universal format list, no every direction, no batch processing and no local processing. The record server processing and retention caveats are unchanged, and both pages are JavaScript applications that returned no text to this tool, so the wording is Calum reading.

  45. 7a053ac

    The AvePDF record, in the protect job

    What it said
    The protect job carried no scope, so a reader could not tell how protection is offered, on which surface, or under what documented limit.
    What it says now
    The job keeps protect and gains the scope browser-based password protection for uploaded PDFs through the lock tool; the documented tool limit is 128 MB and files are processed by the service, from the official lock page read on 2026-10-16 at its localized HTML routes https://avepdf.com/uk/lock-pdf and https://avepdf.com/es/lock-pdf, the English route being JavaScript only. Those pages document adding a password online, uploading from device or cloud, downloading the locked file, the 128 MB tool maximum, Nutrient PDF Protection using GdPicture.NET, and free upload retention for service improvement and model training. The scope infers no cipher strength, no permissions controls, no local processing and no universal size limit. The record retention caveat and the convert scope are unchanged.

  46. 7a053ac

    The axesCheck record, in the view job

    What it said
    The record listed a view job, which asserts that axesCheck can be used to read or view PDFs. Its own page documents checking only: it checks PDF files against the machine-verifiable PDF/UA and WCAG requirements and recommends the additional checks only a human can perform, with no viewing capability claimed.
    What it says now
    The view job is removed and nothing replaced it, so the record no longer claims a viewer it does not have. The accessibility job and its existing scope are unchanged, as is every other field. The one preview-like feature on the page, a screen reader preview supporting visual inspection, belongs to PAC, the separate desktop product, which is also the version that checks a PDF without uploading it, so it is not evidence for axesCheck. Evidence is https://check.axes4.com/en read on 2026-10-16.

  47. 7a053ac

    The axesPDF record, in the secondary edit job

    What it said
    The secondary edit job carried no scope, so the label edit sat bare on a remediation tool whose vendor page draws a deliberate line between its accessibility tools and general PDF editing.
    What it says now
    The job keeps edit and gains the scope checking and correcting tagged PDF accessibility structure and Unicode mappings for PDF/UA and WCAG in the axesPDF desktop app; not general PDF editing, from https://www.axes4.com/en/desktop-software/axespdf read on 2026-10-16. That page says axesPDF corrects PDF/UA errors and lets you edit the entire structure tree, add, move or delete tags and change hierarchies, integrate PAC checks, use a screen reader preview, check and correct Unicode mapping, edit tag properties such as alternative text, set the PDF/UA identifier, define tab order, mark untagged content as artifacts, search the structure tree, repair damaged elements and resolve Form XObjects, and it states that instead of general PDF editing features the tools are tailored to accessible PDF documents. The scope claims no automatic full accessibility and no availability beyond the record existing desktop fields.

  48. 7a053ac

    The Bluebeam Revu record, in the protect job

    What it said
    The protect job carried no scope, so a reader had no statement of which security controls Revu offers or how they apply.
    What it says now
    The job keeps protect and gains the scope password protection, encryption and permission controls for PDFs in Bluebeam Revu, including limits on opening, printing and editing and applying the same security to multiple files at once; plan availability varies by subscription, from https://support.bluebeam.com/online-help/revu20/Content/RevuHelp/Menus/Document/Security/Security%20-%20M.htm read on 2026-10-16. That manual documents that security can restrict access or require a password for modification, the Require Password to Open and Enable Permissions to Limit Printing and Editing options with their Open and Master passwords, an Encryption menu ordered from simplest to most complex, saving and reapplying a security preset, and processing multiple PDFs at once. The scope claims no redaction, no universal plan availability, no cryptographic strength and no parity for browser or mobile companions.

  49. 7a053ac

    The borb record, in the secondary merge job

    What it said
    The merge job carried no scope, so a reader had no statement of what merging means in this library.
    What it says now
    The job keeps merge and gains the scope merging existing PDF documents and manipulating their page trees with the pure-Python borb library, from https://github.com/borb-pdf/borb_examples/blob/master/11/README.md read on 2026-10-16. That chapter is headed Working with existing PDFs and contains a section headed Merging two or more PDFs, saying merging PDFs is a common task, and a section headed Adding and Removing Pages to a PDF, saying borb makes it easy to manipulate PDF structures programmatically. Both sections carry TODO placeholders rather than code, which is recorded rather than hidden. The scope implies no browser or service, no arbitrary Office conversion and no preservation of every form or bookmark feature, and the record licence, telemetry and redaction notes are unchanged.

  50. 7a053ac

    The CamScanner record, in the convert job

    What it said
    The convert job carried no scope, so a reader could not tell which directions the service converts, or that its Office conversion runs on a particular surface with size and password limits.
    What it says now
    The job keeps convert and gains the scope document conversion across the documented image, PDF, text and Office directions in CamScanner Agent; Office conversion runs through the Skill and CLI, Office inputs must be 40 MB or smaller and not password-protected, and same-format or arbitrary cross-Office conversion is excluded, from https://www.camscanner.com/agent-docs/capabilities/conversion/ read on 2026-10-16. That page lists image to Word, Excel, Markdown or one page PDF, PDF to Word, Excel, TXT, Markdown, JPG or a JPG ZIP, TXT to Word, DOC, XLS and PPT to PDF, and legacy DOC, XLS and PPT upgraded to their modern formats, and states the surface, size, password and same-format exclusions. The scope claims no availability in every mobile or web edition, no perfect layout preservation and no merging of existing PDFs. The record server processing caveat is unchanged.

  51. 7a053ac

    The CamScanner record, in the merge job

    What it said
    The record listed a merge job, asserting the service can merge PDFs. The official CamScanner skill denies that outright: its description says it does not support merging existing PDFs into one file, its execution rules say the current CLI cannot merge existing PDFs, its unsupported operations list merging existing PDF or Word or Excel files, its routing table says multiple PDFs to one file is not supported, and its mistakes table lists merge two PDFs as unroutable.
    What it says now
    The merge job is removed. Nothing replaced it. The command image merge-pdf, which merges images into a PDF, was deliberately not treated as evidence for merging PDFs, because the source states that image merge-pdf accepts images only and that converting a PDF to images and recombining them is lossy reconstruction rather than a merge. The convert scope added on 2026-10-16 is unchanged, as is every other field. Evidence is https://raw.githubusercontent.com/CS-CamScanner/camscanner/main/SKILL.md, skill version 1.1.8, read on 2026-10-16.

  52. 7a053ac

    The Datalogics Adobe PDF Library record, in the primary edit job

    What it said
    The primary edit job carried no scope and no recorded vendor evidence, so it rested only on the record own word modifies, which is not independent evidence. The top level product page and the developer documentation hub offered nothing better, only comprehensive PDF manipulation capabilities.
    What it says now
    The job keeps edit and gains the scope modifying existing PDF documents in the officially published .NET samples: optional content layer states, transparency flattening, page labels and low-level PDF objects, from https://raw.githubusercontent.com/datalogics/apdfl-csharp-dotnet-samples/develop/ContentModification/README.md read on 2026-10-16. That official sample category enumerates ChangeLayerConfiguration for optional content group states within a document, FlattenTransparency, PageLabels, ChangeLinkColors, Watermark, MergePDF and PDFObject for low-level PDF objects, all acting on documents that already exist rather than creating, converting or rendering them. The scope does not rest on annotation only samples such as Action, LaunchActions, UnderlinesAndHighlights or AddQRCode, and claims no general text or content editing. It names the published .NET samples rather than asserting a Version 21 guarantee. The primary designation and all other fields are unchanged.

  53. 7a053ac

    The Datalogics Adobe PDF Library record, in the secondary protect job

    What it said
    The protect job carried no scope and no recorded vendor evidence, so it rested on the record own summary word secures. An earlier pass could not settle it because the class page truncated before its member list and the only on-point sample category listed signatures and redaction rather than passwords.
    What it says now
    The job keeps protect and gains the scope password-based PDF security with access permissions and encryption through Document.Secure, including the requirement that the document be fully saved for changes to take effect, from https://docs.datalogics.com/apdfl21/dotnet/apdfl21.0.0plusp1c/class/document/ read on 2026-10-16. That APDFL 21 API page documents void Secure(PermissionFlags permissions, string ownerPassword, string userPassword), overloads taking EncryptionType and a metadata encryption flag, PermissionFlags as an OR of PDF document permissions, optional owner and user passwords, the Standard security handler being assigned if needed, content and metadata encrypted by default with selective encryption through the overload, and the full save requirement. The scope claims no DRM, no certificate or signature based security and no every-security-feature guarantee. The primary edit scope and all other fields are unchanged.

  54. 7a053ac

    The Datalogics Adobe PDF Library record, in the secondary ocr job

    What it said
    The ocr job carried no scope, so a reader could not tell whether recognition produces extracted text or a searchable document layer, or which content it runs on.
    What it says now
    The job keeps ocr and gains the scope OCR of rasterized PDF pages and images through the .NET API, placing recognized text behind the image so the content becomes searchable, from https://raw.githubusercontent.com/datalogics/apdfl-csharp-dotnet-samples/develop/OpticalCharacterRecognition/README.md read on 2026-10-16. That official sample category lists AddTextToDocument, which places recognized text behind the OCR images found on a PDF page, AddTextToImage, which runs OCR on an added image and places the recognized text behind it, and OCRDocument, which runs OCR on the document recognizing text on its rasterized pages. The scope keeps OCR distinct from text extraction, rendering and generic manipulation, names the .NET surface, and makes no accuracy, language or engine claim because the source states none. The primary edit scope, the protect scope and every other field are unchanged.

  55. 7a053ac

    The Datalogics Adobe PDF Library record, in the secondary merge job

    What it said
    The merge job carried no scope, so a reader could not tell whether merging means two documents, a batch, or a mode with preservation guarantees.
    What it says now
    The job keeps merge and gains the scope merging two PDF documents together, through the officially published .NET sample, from https://raw.githubusercontent.com/datalogics/apdfl-csharp-dotnet-samples/develop/ContentModification/README.md read on 2026-10-16, whose MergePDF entry reads Merges two PDF documents together. The source specifies two documents and says nothing further, so the scope claims no batch or arbitrary merging, no bookmark, form or annotation preservation, and no alternative merge mode. The primary edit scope, the protect and ocr scopes and every other field are unchanged.

  56. 7a053ac

    The Datalogics Adobe PDF Library record, in the secondary extract job

    What it said
    The extract job carried no scope, so a reader could not tell what is extracted, and the label could be read as text extraction, which the source does not enumerate.
    What it says now
    The job keeps extract and gains the scope enumerating document information through the officially published .NET samples: bookmarks, document info, layers, paths and XMP metadata, from https://raw.githubusercontent.com/datalogics/apdfl-csharp-dotnet-samples/develop/InformationExtraction/README.md read on 2026-10-16. That official category lists ListBookmarks, ListInfo, ListLayers, ListPaths and Metadata, each described as listing or enumerating what is present in a PDF document. Its entries are all information listing, so the scope claims no text extraction, no OCR, no rendering and no conversion or editing. The primary edit scope, the protect, ocr and merge scopes and every other field are unchanged.

  57. 7a053ac

    The DocuSeal record, in the secondary edit job

    What it said
    The edit job carried no scope, so a reader could take it as general PDF editing, and the presence of a visual form builder does not license that reading.
    What it says now
    The job keeps edit and gains the scope building fillable PDF form templates and placing fields through the WYSIWYG field builder, embedded text field tags or the HTML API; not general PDF content editing, from https://raw.githubusercontent.com/docusealco/docuseal/master/README.md read on 2026-10-16. That readme describes document filling and signing, lists a PDF form fields builder with 12 field types, and lists template creation through an HTML API and through PDF or DOCX with field tags. Nothing there describes editing existing page content, structure, text or annotations, so the scope claims none of it. The primary sign job and every other field are unchanged.

  58. 7a053ac

    The Docling record, in the accessibility job

    What it said
    The record listed an accessibility job, which in this catalogue means producing or remediating tagged, PDF/UA-conformant output. The official issue on that subject is a proposal for discussion, dated 2026-09-02, whose first phase is still open, which states that Docling itself does not write PDF, and which puts validation of PDF/UA-2 conformance out of scope because it stays with veraPDF.
    What it says now
    The accessibility job is removed. Nothing replaced it, and no narrower accessibility capability was scoped, because the only official source found is a proposal rather than a shipped capability. Reading a tagged structure tree and recording which path produced each item is not writing or remediating an accessible PDF. The extract, convert and OCR jobs and the summary are unchanged, as is every other field. Evidence is https://github.com/docling-project/docling/issues/4146, retrieved in full through the GitHub API on 2026-10-16. A measurements row in the same issue mentions our PDF/UA writer, which sits awkwardly beside the statement that Docling does not write PDF; that tension is recorded rather than resolved, and neither reading establishes the job.

  59. 7a053ac

    The Foxit PDF Editor record, in the secondary compress job

    What it said
    The compress job carried no scope, so a reader could not tell which capability or which edition of Foxit PDF Editor performs size reduction.
    What it says now
    The job keeps compress and gains the scope reducing PDF file size through the Optimize PDF File feature in the Foxit PDF Editor desktop editions for Windows and Mac, from two official Foxit sources read on 2026-10-16: the PDF Editor manual for Mac version 2026.2, whose contents list an Optimize PDF File entry, and the user guide page whose title is Optimize Scanned PDF, Foxit PDF Editor for Windows. Both pages truncated before the body of those sections, so the detail about the PDF Optimizer compressing contents, unembedding fonts and clearing redundant objects, and about Reduce File Size covering the current file and multiple files, is attributed to Calum reading rather than claimed as retrieved wording. The scope names only the two editions the sources establish, claims no lossless compression, quality guarantee, batch behaviour or API and mobile support, and the platforms field was deliberately left unchanged.

  60. 7a053ac

    The Foxit PDF Editor record, evidence clarification for the secondary compress job

    What it said
    The ledger entry for the Foxit compress scope recorded that both official pages truncated before the body of their optimize sections, so the scope rested on a page title and a table of contents entry, and the feature detail was attributed to Calum reading rather than quoted.
    What it says now
    The official Foxit PDF Editor manual for Mac 2026.2 is directly retrievable and states, under Optimize PDF Files: With the PDF optimizer, users can update the PDF version and reduce file size by compressing contents, unembedding fonts, and clearing out the data redundancy of PDF objects. It then documents the Reduce File Size command optimizing multiple files at the same time, and its Optimize Scanned PDF section states that Foxit PDF Editor supports size optimization of scanned PDFs. The scope already committed as reducing PDF file size through the Optimize PDF File feature in the Foxit PDF Editor desktop editions for Windows and Mac is therefore supported by the manual text itself, not merely by a title. The edition limitation stands, and no API or mobile claim is added. No source hash is recorded here because the project source hash convention belongs to the claims register rather than to this corrections ledger, so the evidence is the URL and the retrieval date.

  61. 7a053ac

    The Foxit PDF SDK record, in the secondary convert job

    What it said
    The convert job carried no scope, so the label could be read as general format conversion, including Office and HTML paths that this source does not describe.
    What it says now
    The job keeps convert and gains the scope standardising PDF versions and converting to standard-compliant output such as PDF/A, PDF/E and PDF/X through the Compliance add-on, which needs its own resource package and a licence unlock code; not Office, HTML or arbitrary format conversion, from https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/standards-and-compliance/compliance.html read on 2026-10-16. That page says Foxit PDF SDK provides verification and conversion so PDFs meet a specified version or archival, print or engineering-exchange standard, with a table covering PDF version standardisation for 1.3 to 1.7, PDF/A-1a and 1b, 2a, 2b and 2u, 3a, 3b and 3u and from 12.0 PDF/A-4e and 4f, PDF/E-1 and PDF/X-1a, 3, 4 and 4p, each with its own compliance class, plus Preflight analysis, fixup and reporting. It states the Compliance resource package prerequisite and the production unlock code requirement, and notes that PDF/A disables font linking, encryption, JavaScript and audio or video. The scope claims no Office, HTML, DWG or Markdown conversion, no bidirectional Office conversion, no layout preservation, no all-platform claim and no general compliance verification. The conversion overview page was not opened, so no separate-engine or Conversion SDK distinction is asserted here. Summary, platforms, editions and other jobs are unchanged.

  62. 7a053ac

    The Genius Scan record, in the secondary merge job

    What it said
    The merge job carried no scope, so the label could be read as general-purpose PDF merging rather than a step in a mobile scanning workflow.
    What it says now
    The job keeps merge and gains the scope merging documents together in the Genius Scan mobile app as part of the scanning workflow, alongside adding and moving pages of a scanned PDF; not a general-purpose PDF merge tool, from https://thegrizzlylabs.com/genius-scan/ read on 2026-10-16. That page section Create PDFs says a paper document is digitized into a PDF with a text layer added, and that you can then add more pages, move pages around, and merge documents together. It describes the product as a scanner in your pocket with mobile scanning use cases, so the scope names the mobile app. The scope claims no merging of arbitrary imported PDFs, no desktop, web or API availability, no batch merging and no preservation guarantees; the page batch scanning mention is scanning rather than merging and is not used as evidence for batch merging, and the linked separate web app is not asserted. Protect, OCR, the summary, platforms, pricing and every other field are unchanged.

  63. 7a053ac

    The Genius Scan record, in the secondary protect job

    What it said
    The protect job carried no scope, so app or device locking could be mistaken for protecting the PDFs themselves.
    What it says now
    The job keeps protect and gains the scope encrypting the PDFs Genius Scan creates, listed as PDF encryption under PDF creation in the app features; not app lock, access permissions, certificates or DRM, from https://thegrizzlylabs.com/genius-scan/pricing read on 2026-10-16. That page compare table lists Scan to multi-page PDF, PDF encryption and PDF/A support under PDF creation, and separately lists Strong data privacy with GDPR and CCPA, Face ID or Touch ID or Password protection, Auto-deletion and HIPAA support under Privacy and security. The scope rests on the first grouping, keeps app lock out of it, and does not use the privacy, HIPAA or auto-deletion lines as evidence. The compare table has Basic, Ultra and Team columns, but the per-plan tick marks did not survive retrieval, so no plan limitation is asserted. Merge, OCR, the summary, platforms, pricing and every other field are unchanged.

  64. 7a053ac

    The gImageReader record, in the convert job

    What it said
    The convert job carried no scope, so the label could be read as general format conversion rather than PDF generation from recognition output.
    What it says now
    The job keeps convert and gains the scope generating a PDF from hOCR recognition output, either as a reconstructed document or as the source image with an invisible selectable text layer; not arbitrary input or output conversion, from https://raw.githubusercontent.com/manisandro/gImageReader/master/data/manual.html.in read on 2026-10-16. The manual makes the PDF generation a feature, Generate PDF documents from hOCR documents, and says in the hOCR and PDF mode that gImageReader can process hOCR documents further to generate a PDF document for the recognition result. It documents two export modes: a reconstructed PDF with the same layout and graphics as the source, and a PDF with the unmodified source image plus invisible but selectable text overlaid. It also records plain text and ODT export of the hOCR document and an HOCR Batch Export that batch-exports hOCR HTML files to PDF, ODT or text; those are recorded as sourced detail and deliberately left out of the job scope, which makes no batch claim. No arbitrary input or output conversion, PDF to Office, Office to PDF or general PDF editing is claimed. OCR and every other field are unchanged.

  65. 7a053ac

    The Goodnotes record, in the primary view job

    What it said
    The view job carried no scope and no recorded official evidence. Two earlier retrieval attempts failed: the fetch tool returned HTTP 403 with a Cloudflare interstitial, and direct requests to the same host returned HTTP 403 Forbidden, while the marketing pages that did load carried only feature navigation and no viewing wording.
    What it says now
    The job keeps view and gains the scope opening and viewing imported PDFs and Goodnotes documents in the app, documented for Android, Windows and Web; not a standalone PDF viewer. The evidence, retrieved through a source reader by Calum on 2026-10-16 because every direct route was blocked, is the Get Started article, which says it covers basic workflows in Goodnotes for Android, Windows and Web, lists Read Only Mode described as prevent editing in Document View, and under Import a PDF says to select New then Import and to select the PDF from the file picker; and the non Apple device article, which says you can view documents from Goodnotes on a non Apple device by sharing them on the web, backing them up as PDFs, or exporting them in a compatible format, and explains opening the shared link or the PDF copy. The scope names only the platform set the first article states, treats read only mode as a document state rather than proof of a standalone viewer, claims no arbitrary file compatibility, and leaves out the offline access mentioned on the AI information page. The record platform list is unchanged.

  66. 7a053ac

    The Gotenberg record, in the secondary edit job

    What it said
    The edit job carried no scope, so the label could be read as a general PDF editor, which the documented routes do not provide.
    What it says now
    The job keeps edit and gains the scope modifying existing PDFs through documented manipulation routes: writing metadata and bookmarks, flattening form fields and annotations, and stamping or rotating pages; not editing page text or page content, from https://gotenberg.dev/docs/getting-started/routes read on 2026-10-16. That page lists its PDF manipulation routes as merge, split, convert to PDF/A or PDF/UA, read and write metadata, read and write bookmarks, embed attachments, Factur-X or ZUGFeRD e-invoices, flatten form fields and annotations, optimize, watermark, stamp, rotate and encrypt, each with its own route, and says to pick the route matching your task. There is no general edit route and none edits page text, so the scope names only the operations that modify an existing PDF and excludes text or content editing. Merge, split, PDF/A or PDF/UA conversion, encryption and optimization are left to the merge, convert, protect and compress jobs rather than folded in. The summary and every other field are unchanged.

  67. 7a053ac

    The GroupDocs record, in the secondary protect job

    What it said
    The protect job carried no scope, so the label could be read as general document security across the whole GroupDocs family, including encryption strength or certificates the documentation does not describe.
    What it says now
    The job keeps protect and gains the scope adding, updating, removing and checking document passwords through GroupDocs.Merger for .NET, including adding a password to a PDF with permissions; not encryption algorithms, certificates, DRM or signatures, from https://docs.groupdocs.com/merger/net/security-operations/ read on 2026-10-16. That page says GroupDocs.Merger provides a subset of features to manage document security, allowing you to add, update and remove password for documents from different sources, and links guides for adding a document password, adding a password to PDF with permissions, checking password protection, removing and updating a password, under GroupDocs.Merger for .NET. The scope names that component and its .NET documentation rather than every GroupDocs product, runtime or wrapper, and claims no encryption algorithms, certificates, DRM, signatures or access control semantics. Edit, convert, sign, extract, view, merge, the summary and pricing are unchanged.

  68. 7a053ac

    The GroupDocs record, in the secondary merge job

    What it said
    The merge job carried no scope, so the label could be read as arbitrary cross-format merging across every GroupDocs runtime.
    What it says now
    The job keeps merge and gains the scope merging PDF files with GroupDocs.Merger for .NET, joining documents into one and optionally carrying bookmarks; not arbitrary cross-format merging or every GroupDocs runtime, from https://docs.groupdocs.com/merger/net/merge/pdf/ read on 2026-10-16. That page says GroupDocs.Merger allows developers to merge PDF files to organize multiple PDFs into a single document, and that it can combine PDF documents of any size and structure with all text, images, tables, graphs, forms and other content preserved. Its example creates a Merger from the source PDF, calls Join for another PDF and Save for the merged file, and it documents merging with bookmarks through PdfJoinOptions and UseBookmarks. The scope names GroupDocs.Merger and its .NET documentation rather than every runtime, claims no arbitrary cross-format merge, no batch processing and no all-platform coverage, and does not treat the online PDF Merger App as a library capability. Edit, convert, protect, sign, extract, view, the summary and pricing are unchanged.

  69. 7a053ac

    The GroupDocs record, in the secondary extract job

    What it said
    The extract job carried no scope and no recorded component evidence, so it could be read as text extraction across the whole GroupDocs family or as including OCR.
    What it says now
    The job keeps extract and gains the scope extracting text, metadata, images, tables and annotations from PDF documents through GroupDocs.Parser for .NET, using GetText or GetText(pageIndex) for a whole document or a selected page; not OCR or arbitrary document conversion, from https://docs.groupdocs.com/parser/net/extract-text-from-pdf-documents/ read on 2026-10-16. That page says GetText and GetText(pageIndex) extract a text from the entire document or from the selected page, that the steps are to instantiate a Parser, call GetText for a TextReader and read it, and warns that GetText returns null where text extraction is unsupported while an empty PDF yields an empty reader. The same documentation navigation lists metadata, image, table, hyperlink, annotation, form, table of contents and attachment extraction, and places OCR in its own Using OCR section with a connector, so OCR is not claimed. The scope names GroupDocs.Parser and its .NET documentation, claims no arbitrary document conversion, no semantic understanding and no every-format coverage. Edit, convert, protect, sign, view, merge, the summary and pricing are unchanged.

  70. 7a053ac

    The IronPDF record, in the secondary compress job

    What it said
    The compress job carried no scope, so the label could be read as lossless optimization with a guaranteed size outcome.
    What it says now
    The job keeps compress and gains the scope reducing PDF file size through the PdfDocument compression API, which re-encodes embedded images as JPEG at a chosen quality and can strip the PDF structure tree; not lossless compression and no guaranteed size reduction, from https://ironpdf.com/how-to/pdf-compression/ read on 2026-10-16, whose own title is Compress PDF Files in C# - Reduce Image and File Size. A direct request to that page retrieved the article body, and it states that IronPDF compresses PDFs in C# through the CompressAndSaveAs method, which re-encodes embedded images and optionally strips the PDF structure tree to cut file size in a single call, that images are re-encoded as JPEG by default or as 1-bit CCITT Group 4 when opted in, that companion methods CompressPdfToBytes and CompressPdfToStream exist, and that the second argument is the JPEG quality. Calum identified the page and the mechanism; the wording itself was then retrieved and confirmed here, and the examples page was not relied on. The scope claims no lossless compression, no quality preservation, no arbitrary optimization, no all-runtime coverage, no batch behaviour and no guaranteed size reduction. Edit, convert, protect, sign, merge, accessibility, the summary, platforms and pricing are unchanged.

  71. 7a053ac

    The jsPDF record, in the primary convert job

    What it said
    The primary convert job carried no scope, so the label convert could be read as converting other formats into or out of PDF, which the source does not describe.
    What it says now
    The job keeps convert and gains the scope generating PDF documents in JavaScript with the jsPDF API, including creating a document, adding text and saving the PDF; not converting other file formats and not editing existing PDFs, from https://raw.githubusercontent.com/parallax/jsPDF/master/README.md read on 2026-10-16. That readme says jsPDF is a library to generate PDFs in JavaScript, and its usage example imports jsPDF, creates a document with new jsPDF defaulting to A4 portrait millimetres, writes text with doc.text and saves with doc.save, with constructor options for paper size, orientation and units. It documents Node.js use with require and saving into the working directory, other module formats, optional html2canvas and dompurify dependencies for the html method, and custom fonts for UTF-8. The scope claims no arbitrary source-format conversion, no PDF to PDF conversion, no editing of existing PDFs, no batch behaviour, no layout fidelity, no browser-only or Node-only coverage, and no all-framework claim. Every other job and field is unchanged.

  72. 7a053ac

    The LlamaParse record, in the ocr job

    What it said
    The ocr job carried no scope, so the label could be read as a general OCR engine with language and accuracy guarantees, or as local processing.
    What it says now
    The job keeps ocr and gains the scope running OCR automatically over pages with no text layer through the hosted LlamaParse Parse product, including scanned PDFs and image-only documents and images embedded in documents, returning extracted per-page text and markdown; not local or offline processing, from https://developers.llamaindex.ai/llamaparse/parse/features/ocr-and-languages/ read on 2026-10-16. The fetch tool truncated that page to its navigation, so the page and the Parse overview at https://developers.llamaindex.ai/llamaparse/parse/ were retrieved by direct request and their OCR wording checked. The scope keeps the claim tied to the hosted Parse product, claims no handwriting accuracy, no language coverage beyond the source, no general-purpose OCR quality, no local or offline processing, no arbitrary file conversion, no batch guarantees and no all-SDK support. Every other field and job is unchanged.

  73. 7a053ac

    The Master PDF Editor record, in the view job

    What it said
    The view job carried no scope, so the label could be read as a standalone reader product or as covering web and mobile editions.
    What it says now
    The job keeps view and gains the scope opening and viewing PDF documents in the Master PDF Editor desktop application, with opened files shown as tabs in the view area; not a standalone reader and not web or mobile, from https://code-industry.net/masterpdfeditor-help/opening-and-saving-pdf-files/ read on 2026-10-16. That page says Master PDF Editor is fully compatible with PDF specification 1.7 and supports dynamic XFA forms, allowing it to open and view files created in other PDF editors, lists opening through File then Open or Ctrl+O, the Open File toolbar button, drag and drop, and an Explorer context menu entry, notes files can arrive from the Internet, a local network or e-mail, and states that opened files are displayed as tabs in the view area. The companion document navigation page was not opened here, so page navigation beyond the opening page statements is credited to Calum reading rather than asserted. The scope claims no standalone reader product, no web or mobile support, no arbitrary file types, no rendering fidelity, no annotations or editing, no password handling and no all-OS coverage. Compress and every other field are unchanged.

  74. 7a053ac

    The OCRmyPDF record, in the compress job

    What it said
    The compress job carried no scope, so the label could be read as arbitrary PDF optimization guaranteeing a smaller file.
    What it says now
    The job keeps compress and gains the scope reducing PDF size through OCRmyPDF image-oriented optimizer, which runs after a successful OCR: lossless optimisation at the default level and optional lossy optimisation at higher levels, with the output size not guaranteed to decrease, from https://ocrmypdf.readthedocs.io/en/latest/optimizer.html read on 2026-10-16. That page says OCRmyPDF includes an image-oriented PDF optimizer, that by default it runs with safe settings aiming to improve compression at no loss of quality, that higher levels may apply lossy optimizations, and that optimization occurs after OCR and only if OCR succeeded. It documents optimize levels 0 to 3, with level 1 the default enabling lossless image and object optimizations and levels 2 and 3 adding lossy optimization and color quantization and then lower image quality, notes some lossless replacements always occur and cannot be disabled, and warns that OCRmyPDF might still increase the overall file size because it must embed the recognized text. The scope claims no arbitrary PDF optimization, no lossless behaviour at every level, no quality preservation at higher levels, no batch behaviour and no size reduction guarantee. Every other field and job is unchanged.

  75. 7a053ac

    The PDFsam record, in the convert job

    What it said
    The convert job carried no scope, so the label could be read as a capability of the family, including the free open source Basic edition, rather than of one paid edition.
    What it says now
    The job keeps convert and gains the scope converting PDFs to Office formats and creating PDFs from 300+ formats with PDFsam Enhanced on Windows, where the free view and create capability requires registration; not PDFsam Basic or Visual, from https://pdfsam.org/downloads/ read on 2026-10-16. That page describes PDFsam Enhanced 7.0.0 as a complete PDF editor for Windows that can edit text and images, convert to Office formats, fill and create forms, annotate and sign, free to view and create PDFs from 300+ formats, registration required, paid to edit, with Windows as its only listed platform. PDFsam Basic is described as the free open source app to split, merge, extract, rotate and mix PDF files, and PDFsam Visual as a toolbox of more than 25 tools to organise pages, compress, crop, split by text, flatten and clean metadata, so neither is attributed conversion. The page also says all three apps run offline and files are never uploaded, which is recorded rather than folded into the scope, because the scope claims no offline conversion. No format coverage beyond the 300+ figure, no Office format list, no round-trip fidelity, no batch behaviour and no API support are claimed. Every other field and job is unchanged.

  76. 7a053ac

    The PDF24 record, in the convert job

    What it said
    The convert job carried no scope, so the label could be read as covering the whole PDF24 tool set, or as offline conversion through the Windows Creator app.
    What it says now
    The job keeps convert and gains the scope converting common Office and image files to PDF and converting PDFs back to editable DOCX and other formats in PDF24 online PDF Converter, where the conversion is done on PDF24 servers; not a claim about every format in the wider PDF24 tool set, from https://tools.pdf24.org/en/pdf-converter read on 2026-10-16. That page subtitle says it turns Word, Excel, PowerPoint and image files into PDF or changes PDFs back into editable DOCX and other formats in one tool, offers a Convert to PDF and a Convert PDF to direction, lists frequently used input types DOC, DOCX, XLS, XLSX, PPT, PPTX, ODT, ODG, ODS, ODP, JPG, PNG and TIFF, states under No installation necessary that the conversion is done on our servers, says the online converter works under all common operating systems and browsers, and in its FAQ points to the free PDF24 Creator desktop program as the offline solution. The scope is therefore online and server processed, matching the record split between web tools and the Creator app, and claims no every-format coverage, no arbitrary conversion, no round-trip fidelity, no batch behaviour, no offline conversion, no API support and no format list beyond the page examples. Every other field and job is unchanged.

  77. 7a053ac

    The PDF Expert record, in the merge job

    What it said
    The merge job carried no scope, so the label could be read as general PDF merging including web or Windows, and as an entire-document only operation.
    What it says now
    The job keeps merge and gains the scope merging entire PDF documents or selected pages from multiple PDFs in PDF Expert on Apple devices; the vendor states the app is not available on Windows and the web offering is a separate product, from https://pdfexpert.com/features/combine-pdf-mac read on 2026-10-16. That page says you can combine multiple PDFs into one with PDF Expert, the go-to PDF app for Mac, and that its tools allow you to rearrange, extract, delete, rotate pages or merge entire PDF documents. Its FAQ documents File then Merge Files or drag and drop followed by Merge, says you can merge two or multiple PDFs, and states that you can merge specific pages from multiple PDFs, not just entire documents, by selecting the pages to include from each. The page also says there are no limits on the number of files you can combine, which is recorded rather than folded into the scope; and it states the app is not available on Windows, with the browser offering being PDF Copilot rather than the app. The scope claims no arbitrary file types, no web support, no Windows or Android support, no API or library behaviour, no lossless preservation wording and no subscription entitlement. Compress, platforms, the summary and every other field are unchanged.

  78. 7a053ac

    The PDF.js record, in the convert job

    What it said
    The record listed a convert job, which asserted a document conversion capability. The official project description says PDF.js is a general-purpose, web standards-based platform for parsing and rendering PDFs, and nothing in the record or in that description establishes conversion.
    What it says now
    The convert job is removed and nothing replaced it. Rendering a page to a canvas is display and getting other information out of a document is extraction, so neither is treated as conversion, and no conversion to another file format or creation of a transformed PDF is claimed. The Getting Started page describes the display layer API as rendering PDFs and getting other information out of a document, and the viewer as the UI for the PDF viewer in Firefox and browser extensions. The primary view job, edit, the summary, platforms and every other field are unchanged. Evidence is https://mozilla.github.io/pdf.js/ read on 2026-10-16.

  79. 7a053ac

    The ReportLab record, in the edit job

    What it said
    The record listed an edit job, which asserted editing existing PDF content. The official introduction describes ReportLab as a software library that lets you directly create documents in Adobe Portable Document Format using the Python programming language, and the component that touches existing PDFs, PageCatcher, is a commercial route to importing and reusing existing pages rather than editing their content.
    What it says now
    The edit job is removed and nothing replaced it. ReportLab generation and PageCatcher page reuse are not editing, so neither is presented as one, and no narrow edit scope was forced. The record already separates the open source toolkit generation from commercial page reuse. The primary convert job and its scope, merge, protect, accessibility, the summary and every other field are unchanged. Evidence is https://docs.reportlab.com/reportlab/userguide/ch1_intro/ and https://docs.reportlab.com/pagecatcher/ read on 2026-10-16, retrieved by direct request because the fetch tool truncated the introduction page to its navigation.

  80. 7a053ac

    The PDF Studio record, in the secondary merge job

    What it said
    The merge job carried no scope, so the label could be read as accepting any file type or as a preservation guarantee.
    What it says now
    The job keeps merge and gains the scope merging supported input files, including PDFs, into a single PDF in PDF Studio through the Add Files or Add Folder workflow, with optional bookmarks for each merged document and results opened, appended or saved; not arbitrary formats beyond the documented Supported File Types list, from https://www.qoppa.com/files/pdfstudio/guide/merge-convert-multiple-files-to-pdf.htm read on 2026-10-16. That page says PDF Studio can merge or convert multiple files at once into a single PDF and that only Supported File Types can be merged or converted, documents three routes to the command, the Add Files and Add Folder selection, a bookmark per merged document option, open or save results, up to four passwords for password-protected PDFs, page size and fitting settings, and result options of open new merged document, append to current document or save as. It notes all files are merged in a single action. The scope claims no format coverage beyond the guide supported file types page, no lossless preservation wording, no API or library behaviour, no batch claim beyond the single action the guide describes, and no complete supported-type list. The primary edit job and every other field are unchanged.

  81. 7a053ac

    The PDF Studio record, in the secondary protect job

    What it said
    The protect job carried no scope, so the label could be read as certificate based security, DRM or a specific encryption strength.
    What it says now
    The job keeps protect and gains the scope setting, changing or removing a PDF open password and permissions password and its permission restrictions in PDF Studio Document Security settings, the change applying only once the document is saved; not certificates, DRM or specific encryption algorithms, from https://www.qoppa.com/files/pdfstudio/guide/document-security.html read on 2026-10-16. That page says PDF Studio can set or change passwords and permissions, that permissions restrict how a document may be modified and accessed, and that existing permissions can be edited or removed. It defines the open or user password as required to open the document and the permissions or owner password as required to change permissions and passwords, lists permissions covering high resolution printing, printing, extraction, extraction for accessibility, document changes, assembly, annotations and comments and form filling and signing with their automatic cascades, and documents editing, creating and removing settings through File, Properties, Security, Edit and Save, noting the settings apply only after saving and that secured files show SECURED after the name. The scope claims no certificate signing, no DRM, no encryption algorithm or strength, no batch processing, no lossless preservation and no platform or edition entitlement. The primary edit job, the merge scope and every other field are unchanged.

  82. 7a053ac

    The PDF Studio record, in the secondary compress job

    What it said
    The compress job carried no scope, so the label could be read as arbitrary optimization or as promising a smaller file.
    What it says now
    The job keeps compress and gains the scope reducing PDF file size by optimizing specific parts of the PDF through PDF Studio documented Optimizing PDFs tool, from https://www.qoppa.com/files/pdfstudio/guide/edit-and-manipulate-pdfs.htm read on 2026-10-16. That page lists Optimizing PDFs as reducing the file size of the PDF by optimizing specific parts of the PDF, among tools for content editing, redaction, modifying and splitting pages, bookmarks, links, destinations, headers and footers, watermarks and OCR. The scope claims no lossless compression, no quality preservation, no guaranteed size reduction, no image re-encoding detail, no arbitrary optimization, no batch behaviour, no API support and no plan or platform entitlement. The primary edit job remains unscoped, the merge and protect scopes are intact, and every other field is unchanged. The same page lists Content Editing, described as editing text, images and content within the PDF, and Redaction, described as securely redacting text and images, which are recorded here as findings for future passes rather than acted on.

  83. 7a053ac

    The PDF Studio record, in the accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as a standalone accessibility checker or as guaranteeing conformance.
    What it says now
    The job keeps accessibility and gains the scope checking a document against PDF/UA standards with PDF Studio PDF/UA Validation, and creating or editing document tags, alternative text, field tab order, form field tooltips and document reading properties for screen readers; the guide describes more accessible PDFs rather than a conformance guarantee, from https://www.qoppa.com/files/pdfstudio/guide/creating-accessible-pdfs.htm read on 2026-10-16. That page says the features allow you to create more accessible PDFs compatible with screen readers and other assistive devices, lists PDF/UA Validation as checking the document against PDF/UA standards, and lists the tagging tools, alternative text, tab order, field tooltips, general properties for screen readers and advanced reading options such as binding and language. Several of those entries carry a pro marker in the guide, which is recorded as the observed edition signal rather than asserted as an entitlement. The scope claims no conformance guarantee, no automatic fixing of every issue, no standalone checker, no WCAG or ADA coverage beyond the documented PDF/UA language and no every platform or edition entitlement. The companion validation page was not opened here, so its preflight and export detail is credited to Calum reading, and OCR is left to the ocr job. The primary edit job stays unscoped and the settled merge, protect and compress scopes are intact.

  84. 7a053ac

    The Ghostscript record, in the primary compress job

    What it said
    The compress job carried no scope, so it could be read as lossless optimization that preserves everything in the file.
    What it says now
    The job keeps compress and gains the scope reducing output PDF size when Ghostscript pdfwrite device rewrites a PDF, using documented controls such as duplicate-image detection, compressed fonts and streams, and Flate or JPEG image parameters; pdfwrite creates a new PDF and may not carry across non-marking information such as comments, from https://ghostscript.readthedocs.io/en/master/VectorDevices.html read on 2026-10-16. That page says pdfwrite creates a new PDF rather than modifying the original, that non-marking content such as hyperlinks, bookmarks and comments will normally not be present and comments are specifically not preserved, and that pdfwrite performs well where a modified file is the intention, giving reduced image resolution and changed colour space as examples. It documents CompressFonts, CompressStreams, DetectDuplicateImages, MaxInlineImageSize with its note about reducing the size of the generated PDF, Optimize, CompressPages, UseFlateCompression, PassThroughJPEGImages and PassThroughJPXImages, the JPEG image dictionary keys, and the PDFSETTINGS presets with the warning that presets alter the input and may reduce quality. the page truncated before the duplicate image sentence finished and before the optimization section; the XRef and object stream wording is quoted directly from the Ghostscript 10.03.0 release notes, which state that as of that release pdfwrite creates PDF files with XRef streams and ObjStm streams and that this can result in considerably smaller PDF output files. The scope claims no lossless compression, no guaranteed size reduction, no preservation of comments, bookmarks or annotations, no arbitrary optimization, no batch behaviour, no API support and no quality guarantee. Convert, OCR, the summary and platforms are unchanged.

  85. 7a053ac

    The Ghostscript record, in the secondary convert job

    What it said
    The convert job carried no scope, so it could be read as arbitrary or round-trip format conversion.
    What it says now
    The job keeps convert and gains the scope converting supported PostScript, PDF, XPS, PCL or PXL input to a newly written PDF through Ghostscript pdfwrite device, where the output should look the same as the input but is not the same file and non-marking content such as comments may not carry across, from https://ghostscript.readthedocs.io/en/master/VectorDevices.html read on 2026-10-16. That page lists pdfwrite, ps2write and eps2write among the high level devices, says the input is PostScript, PDF, XPS, PCL or PXL handled by an appropriate interpreter, and says those devices create PDF or PostScript whose visual appearance should match as closely as possible the appearance of the original input. The same page explains the caveat the scope carries: the output should be visually the same because it makes the same marks, but it is not the same as the original input, a new PDF file is created, and non-marking content such as hyperlinks, bookmarks and comments is normally not present, with comments specifically not preserved. The scope claims no arbitrary format conversion, no Office conversion, no round-trip fidelity, no preservation of comments, bookmarks or annotations, no batch behaviour, no API support and no guaranteed visual match. The primary compress scope and every other field are unchanged. This pass reused the page read earlier the same session for the compress job.

  86. 7a053ac

    The Ghostscript record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as OCR present in every Ghostscript build, or as a general OCR engine with searchable quality guarantees.
    What it says now
    The job keeps ocr and gains the scope OCR through the pdfocr8, pdfocr24 and pdfocr32 devices in Ghostscript builds that include the Tesseract engine, which render input to an image, OCR it and wrap the image as a PDF with the recognised text added as invisible text; not available in every build, from https://ghostscript.readthedocs.io/en/gs10.03.0/News.html read on 2026-10-16. That page says the capability to build with the Tesseract OCR engine was added in 9.53.0, that in such a build the pdfocr devices render the output file to an image, OCR that image and wrap it as a PDF with the OCR text included as invisible text, text rendering mode 3. It also states that only Tesseract from source in their release packages is supported, not linking to Tesseract or Leptonica shared libraries, and carries a security note about a vulnerability in how Ghostscript called tesseract for the OCR devices. The scope claims no availability in every build, no accuracy or language coverage, no searchable quality guarantee, no arbitrary OCR inputs, no batch behaviour, no API support and no structure preservation claim. The settled compress and convert scopes and every other field are unchanged. The same release notes give direct official wording for the XRef and ObjStm output streams that reduce PDF size, which the compress ledger had attributed to reading; that is noted for a later evidence pass.

  87. 7a053ac

    The iText record, in the convert job

    What it said
    The convert job carried no scope and no prose word, so the mapping rested on nothing recorded.
    What it says now
    The job keeps convert and gains the scope converting HTML and CSS to PDF through the pdfHTML add-on of the iText PDF SDK; not arbitrary format conversion and not part of the base library, from https://itextpdf.com/products/pdfhtml, which resolves to the PDF HTML product page, read on 2026-10-16. That page is titled Convert HTML and CSS to rich, smart PDF documents and its navigation lists pdfHTML as an add-on under the iText PDF SDK, labelled Convert HTML and CSS to PDF, alongside pdfCalligraph, pdfOCR, pdfOptimizer, pdfSweep and pdfXFA, with the Apryse SDK entries listed separately as another vendor products. The scope claims no arbitrary or Office format conversion, no round-trip fidelity and no base-library capability, because the page presents pdfHTML as an add-on. The protect, merge, accessibility and sign jobs on this record are untouched, as are the summary, platforms and every other field.

  88. 7a053ac

    The iText record, in the ocr job

    What it said
    The ocr job carried no scope and no prose word, so the mapping rested on nothing recorded, and the only mention of it was a navigation label.
    What it says now
    The job keeps ocr and gains the scope OCR text recognition in scanned documents, PDFs and images through the pdfOCR add-on of the iText PDF SDK; not part of the base library, from https://itextpdf.com/products/pdf-ocr-text-recognition read on 2026-10-16. The page own title reads OCR text in scanned documents, PDFs and images with pdfOCR, which is a page level statement rather than a navigation mention, and the navigation labels pdfOCR as PDF OCR text recognition among the iText PDF SDK add-ons. The scope claims no base-library OCR, no accuracy or language coverage, no searchable quality guarantee, no arbitrary input handling and no batch behaviour. The record existing add-on licensing and Java and .NET caveats are untouched, as are the convert scope from the pdfHTML page, the other jobs, the summary and platforms.

  89. 7a053ac

    The iText record, in the compress job

    What it said
    The compress job carried no scope and no prose word, so the mapping rested on nothing recorded.
    What it says now
    The job keeps compress and gains the scope automatically optimising PDFs, including in bulk, through the pdfOptimizer add-on of the iText PDF SDK; not part of the base library, from https://itextpdf.com/products/compress-pdf-pdfoptimizer read on 2026-10-16. The page own title reads Automatically optimize PDFs in bulk with pdfOptimizer, which is a page level statement rather than a navigation mention, and the navigation lists pdfOptimizer among the iText PDF SDK add-ons beside pdfCalligraph, pdfHTML, pdfOCR, pdfSweep and pdfXFA, with the Apryse SDK entries listed separately as another vendor products. The bulk wording is the page own, so it is stated rather than excluded. The scope claims no lossless behaviour, no quality or size guarantee, no arbitrary input support and no base-library capability. The record add-on licensing and Java and .NET caveats are untouched, as are the convert and ocr scopes and the other jobs, the summary and platforms.

  90. 7a053ac

    The iText record, in the merge job

    What it said
    The merge job carried no scope and no prose word, so the mapping rested on nothing recorded, and every iText page reached before this had truncated to its navigation.
    What it says now
    The job keeps merge and gains the scope merging multiple PDFs into one as part of the PDF manipulation features of iText Core/Community for Java and .NET, from https://raw.githubusercontent.com/itext/itext-java/develop/README.md read on 2026-10-16. That readme says iText Core/Community allows you to create, adapt, inspect and maintain PDF documents and is also available for .NET, and lists under Core library the features PDF creation with a layout engine, PDF manipulation described as merging multiple PDFs into one and adding new content, PDF digital signing, PDF form creation and manipulation, working with PDF/A and PDF/UA documents, FIPS-compliant cryptography, barcode generation and SVG support, with a separate Addons list covering conversion, OCR, redaction, character sets, optimization, XFA flattening and debugging. Merging therefore sits in the base library rather than an add-on, which the scope states. The convert, ocr and compress scopes and every other field are unchanged. The same feature list names FIPS-compliant cryptography and working with PDF/UA documents, which are recorded as findings for the protect and accessibility jobs rather than acted on here.

  91. 7a053ac

    The iText record, in the accessibility job

    What it said
    The accessibility job carried no scope and no prose word, and the only lead was the readme phrase working with PDF/UA documents, which on its own would not have justified any accessibility claim.
    What it says now
    The job keeps accessibility and gains the scope creating PDF/UA documents and running PDF/UA-1 conformance checks through the PdfUADocument and PdfUAConfig APIs of iText Core, including setting the document as tagged and accessibility properties on form fields; the checks are programmatic support for many, not all, Matterhorn Protocol checkpoints, from https://kb.itextpdf.com/itext/increased-pdf-ua-support read on 2026-10-16. That page describes the iText Core 8.0.4 release increased PDF/UA support, says PDF/UA is an ISO standardised specification containing the accessibility requirements of PDF software and documents, says the Matterhorn Protocol contains the criteria for PDF/UA-1 conformance across over thirty checkpoints, and says iText used it to develop programmatic support for many of the conformance checks, exposing PdfUAConformanceException and PdfUAExceptionMessageConstants. It documents the new PdfUADocument and PdfUAConfig APIs, notes that earlier versions required setting the document as tagged manually with no internal checking, and describes IAccessibleElement on form fields for roles, languages and other PDF/UA information. The scope claims no conformance guarantee, no automatic remediation, no certification, no WCAG or ADA coverage beyond PDF/UA-1, and no add-on framing, since these APIs are in iText Core. The protect job remains unresolved and every other field is unchanged.

  92. 7a053ac

    The iText record, in the protect job

    What it said
    The protect job carried no scope and no prose word, and the two pages opened for it returned only navigation shells, so the mapping rested on nothing recorded. It was left unresolved rather than removed on absence.
    What it says now
    The job keeps protect and gains the scope encrypting PDFs with a user and or owner password and decrypting a PDF with its owner password, as shown in the iText Core security examples for Java and .NET, from https://kb.itextpdf.com/itext/encrypting-decrypting-pdfs read on 2026-10-16. That page presents worked security examples answering questions such as how to decrypt a PDF document with the owner password, and lists encryptpdf, encryptpdfwithoutuserpassword, decryptpdf, decryptpdf2 and decryptpdf3, each with a Java and a .NET C# source file, under the sandbox security package, with hello.pdf and hello_encrypted.pdf as resources. The scope claims no certificate encryption, no AES-GCM, no DRM, no algorithm or strength and no signing, although the site documents certificate encryption and AES-GCM on separate pages, because this page does not state them. The accessibility scope and the convert, ocr, compress and merge scopes are unchanged, as are the summary and platforms.

  93. 7a053ac

    The iText record, in the sign job

    What it said
    The sign job carried no scope and no prose word, so the mapping rested on nothing recorded.
    What it says now
    The job keeps sign and gains the scope digitally signing PDFs through iText high level signing API, with examples for Java and .NET, from https://kb.itextpdf.com/itext/digital-signing-with-itext read on 2026-10-16. That page says signing a PDF with early iText versions required understanding PDF signatures at byte level, that a higher level signing API was introduced in the 5.3.x versions requiring only the knowledge needed for arbitrary digital signature tasks, that iText 7 copied that API and iText 8 improved it, and that the series describes the high level signing API with examples for both Java and .NET C#. Its parts cover overview and simple cases, proprietary APIs and custom providers, PKCS#11, appearances and the CSC API. The scope claims no certificate profiles, no PAdES levels, no trust validation, no timestamping, no cryptographic algorithms despite the page noting iText 8 supports additional ones, and no legal validity. The six scopes already settled on this record and every other field are unchanged.

  94. 7a053ac

    The iText record, in the primary edit job

    What it said
    The primary edit job carried no scope and no prose word, so the mapping rested on nothing recorded.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing an existing PDF by opening it with a reader and a writer and adding additional content such as a new page, a paragraph or an image, with examples for Java and .NET, from https://kb.itextpdf.com/itext/creating-and-editing-pdf-2-0-documents read on 2026-10-16. That page says iText is more than capable of editing an existing PDF 2.0 document and doing things like adding a signature or adding additional content, and demonstrates it with Java and C# code that opens an existing document using a PdfReader and a PdfWriter, adds a new page, and adds a paragraph and an image. The scope claims no arbitrary visual editing, no coverage of every page, text or image edit, no graphical interface and no claim to all PDF 2.0 features, although the page names AES-256 encryption, Unicode passwords and associated files among the specification additions. Its consuming section shows reading a PDF through PdfReader and extracting associated files from pages, which is recorded as a finding for the extract job rather than acted on. The seven earlier scopes and every other field are unchanged.

  95. 7a053ac

    The iText record, in the extract job

    What it said
    The extract job carried no scope and no prose word, so the label could be read as covering tables, images, attachments, metadata or arbitrary object extraction.
    What it says now
    The job keeps extract and gains the scope extracting text from PDFs using text extraction strategies and location extraction strategies, as demonstrated by the Parsing PDFs examples in Java and .NET, from https://kb.itextpdf.com/itext/parsing-pdfs read on 2026-10-16. That page presents examples in answer to a question about using a text extraction strategy after applying a location extraction strategy, and lists parsecustom and parseczech with both Java and .NET C# sources under the sandbox parse package, with czech.pdf and nameddestinations.pdf as resources. The scope claims no table, image, attachment, associated file, metadata or arbitrary object extraction, since the page states none of them. The view job remains unresolved because parsing and extraction are not viewing. The eight earlier scopes and every other field are unchanged.

  96. 7a053ac

    The iText record, in the view job

    What it said
    The view job rested on a prose-only signal with no vendor page read, so the label could be read as a reader application, a graphical interface or general on-screen viewing.
    What it says now
    The job keeps view and gains the scope rendering PDF pages to image output such as JPG, PNG or an in-memory image through the licensed pdfRender add-on, for example to preview a page as an image; this is rasterising to an image, not on-screen display, and is not part of the base library, from https://kb.itextpdf.com/itext/pdfrender-bufferedimage-as-output-format read on 2026-10-16. That page says the pdfRender add-on enables you to easily display, in parentheses render, any PDF as an image, that PDF pages could previously be converted to JPG, PNG or other image formats, and that BufferedImage output was added so an in memory image can be created when converting from PDF. It demonstrates PdfToImageRenderer.renderPdf with a page range and an image ready listener, and shows a licence key being loaded, which is why the scope says licensed add-on. The boundary is the page own clarification that editing the image before displaying it is illustrated by previewing a page in black and white and that displaying it in the example simply means adding it to a PDF file. The scope claims no graphical interface, no reader or viewer application, no on-screen display, no general viewing and no base library capability. The nine earlier scopes and every other field are unchanged. The parsing article was deliberately not used as evidence here, because parsing and extraction are not viewing.

  97. 7a053ac

    The ABBYY FineReader PDF record, in the primary ocr job

    What it said
    The ocr job rested on the record own prose, a prose-only candidate signal, and the only ABBYY recognition documentation reachable previously was for FineReader Engine, an SDK product, which did not match this installed application record.
    What it says now
    The job keeps ocr, remains primary, and gains the scope document recognition of imported images in an OCR project, converting them to text, with automatic recognition of new pages and the Recognize All Pages or Recognize Selected Pages commands, in ABBYY FineReader PDF, from https://help.abbyy.com/finereader/15mac/user_guide/rec_general/ read on 2026-10-16. That page is the Document recognition page of the ABBYY FineReader 15 for Mac User Guide, which is product documentation rather than the Engine SDK. It says images imported to an OCR project need to be recognized, that is converted to text, that the product analyzes the document structure and marks text, picture and table areas before recognition, that all new pages are recognized automatically by default, and that recognition should be run manually after markup, text property, table structure or language changes. It documents the recognize all pages and recognize selected pages commands and notes that export triggers recognition of any pages not yet recognized. The scope claims no language counts or list, no accuracy figure, no searchable PDF or PDF/A output, no batch behaviour, no platform or edition breadth, and no Engine SDK capability. The convert, edit, protect and sign jobs and every other field are unchanged, and the guide being the Mac edition is noted rather than generalised.

  98. 7a053ac

    The ABBYY FineReader PDF record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as arbitrary format conversion, batch conversion or PDF/A output.
    What it says now
    The job keeps convert and gains the scope quick conversion of recognized text to searchable PDF, image-only PDF, DOCX, XLSX, PPTX, EPUB, FB2, TXT, HTML, RTF, ODT or CSV, in ABBYY FineReader PDF, from https://help.abbyy.com/finereader/15mac/user_guide/qs_general/ read on 2026-10-16. That page is the Quick conversion mode page of the FineReader 15 for Mac User Guide and says the mode lets you convert recognized text to a specific format in just a few clicks, listing exactly those formats, mentioning that image preprocessing settings can be changed, and pointing at advanced conversion mode for editing the document layout to improve quality. The scope names the vendor list rather than claiming arbitrary format support, and claims no round-trip fidelity, no batch behaviour and no PDF/A output, which the list does not include. The guide being the Mac edition is noted rather than generalised. The OCR scope, the other jobs and every other field are unchanged.

  99. 7a053ac

    The ABBYY FineReader PDF record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as certificate encryption, DRM, algorithm level claims or protection beyond exported PDF documents.
    What it says now
    The job keeps protect and gains the scope protecting exported PDF documents with passwords, where different passwords can be required for opening a document and for printing and copying its contents, in ABBYY FineReader PDF, from https://help.abbyy.com/finereader/15mac/user_guide/save_security/ read on 2026-10-16. That page is the Protecting your exported PDF document using a password page of the FineReader 15 for Mac User Guide and says the product allows a password to protect exported PDF documents and that different passwords can be required for opening a PDF and for printing and copying its contents. It documents the export flow through the File menu or Export button, the PDF tab, export settings, the file name and storage location, three saving options, then Security Settings where the passwords are specified. The scope uses the page own wording rather than the word permissions, which the page does not use, and claims no certificate encryption, no encryption algorithm or strength, no DRM, no protection beyond exported PDFs and no platform or edition breadth, the guide being the Mac edition. The OCR and convert scopes and every other field are unchanged.

  100. 7a053ac

    The ABBYY FineReader PDF record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as arbitrary editing of PDF content, page insertion or deletion, annotation editing or broad graphical editing.
    What it says now
    The job keeps edit and gains the scope editing the automatically generated document layout in the OCR Editor, using the area tools and Area menu commands to add, delete, resize, move or change the type of areas and to mark out a page, in ABBYY FineReader PDF, from https://help.abbyy.com/finereader/15mac/user_guide/areaedit/ read on 2026-10-16. That page is the Editing a document layout page of the FineReader 15 for Mac User Guide and says that to edit the automatically generated layout for a document you can use the special area tools or the Area app menu commands, with sections for adding, deleting, resizing and moving areas, adding or deleting a rectangular part of an area, changing the area type and automatically marking out a page. It notes that changing the document layout, text properties or table structure requires launching document recognition and that an unmarked part of a page is not recreated in the result document. The scope claims no arbitrary PDF content editing, no page insertion or deletion, no annotations, no PDF text editing and no graphical interface breadth. The guide navigation names text properties, table structure and the Image Editor, but those pages were not opened here and their capabilities are deliberately not claimed. The OCR, convert and protect scopes and every other field are unchanged.

  101. 7a053ac

    The AccessibilityNow record, in the primary accessibility job

    What it said
    The accessibility job rested on the record own prose, a prose-only candidate signal, so it could also be read as certification, automatic fixing of every issue or standards breadth the vendor does not state.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope remediating PDF accessibility by tagging and tag-finishing, starting from an untagged PDF, with settings to match the document structure and support for PDF/UA, WCAG and HHS accessibility standards, in AccessibilityNow Remediate, from https://crawfordtech.com/products/accessibilitynow-remediate/ and https://crawfordtech.com/products/accessibilitynow/ read on 2026-10-16 by direct request because the fetch tool returned only the page title. Those pages describe Remediate as a standalone desktop PDF remediation tool, purpose-built to provide tagging or tag-finishing either standalone or as a final step for already tagged files, and as a graphical user interface tool that starting from an untagged PDF lets you tag documents to make them accessible, with tagging covering read order, paragraphs, heading control, lists, URLs and tables, autosensing tagging elements, customizable settings to match document structure, and support for PDF/UA, WCAG and HHS document accessibility standards. The scope claims no certification, no automatic fixing of every issue, no ADA or Section 508 breadth beyond the HHS standard named, no accuracy or completeness claim, and no platform claim beyond the vendor wording, which frames it as a desktop tool while the record platforms field says web and self host and was left unchanged. The convert job and every other field are unchanged.

  102. 7a053ac

    The AccessibilityNow record, whole file, checked after its accessibility scope landed

    What it said
    A probe reported that the record summary and other fields differed from the previous commit, which if real would have been an unintended content change made alongside a job scope.
    What it says now
    Proved to be a false alarm, with no content change needed. The raw diff of src/content/tools/accessibilitynow.json against the content parent of b40aa97 adds exactly one line, the scope on the accessibility job, and touches nothing else. The comparison that reported differences read the UTF-8 file through PowerShell without the encoding flag, so an em dash was displayed as mojibake and the strings compared unequal; the same artefact affected an earlier probe on the same record. Restoring the file from the parent and reapplying only the scope produced nothing to commit, which confirms the committed file already held exactly the intended change. Sources for the scope itself are the Crawford Technologies AccessibilityNow Remediate product pages, accessed 2026-10-16. This is the AccessibilityNow integrity proof. No product record and no inventory semantics were altered by this entry.

  103. 7a053ac

    The AccessibilityNow record, in the convert job

    What it said
    The convert job rested on the record own tagline, a prose-only candidate signal, so it could be read as arbitrary format conversion or Braille output the vendor does not state.
    What it says now
    The job keeps convert and gains the scope converting documents such as marketing literature, letters, books and manuals into print-ready Accessible PDF, and converting documents into Large Print, in AccessibilityNow, from https://crawfordtech.com/products/accessibilitynow-remediate/ and https://crawfordtech.com/products/accessibilitynow/ re-retrieved by direct request on 2026-10-16. Those pages state that documents such as marketing literature, letters, books, manuals and other communications can be converted into print-ready Accessible PDF, and that documents can be converted into Large Print for printing or digital delivery; a further statement has the AccessibilityNow Gateway retrieving, converting and presenting information from archive systems and ECM repositories for multi-channel delivery. The scope names only the two format-specific conversions and does not claim Braille conversion even though the record tagline mentions it, because the vendor text defines Braille rather than documenting a conversion into it. It also claims no arbitrary format support, no round-trip fidelity, no batch behaviour, no input format list and no platform or edition breadth, and the gateway retrieval statement is not folded in. The accessibility scope and every other field are unchanged.

  104. 7a053ac

    The AccessibilityNow record, in the deployment note

    What it said
    The record carried a flagged contradiction between its platforms field, web and self_hosted, and the vendor describing AccessibilityNow Remediate as a standalone desktop and graphical user interface tool. The deployment note also said no further detail was read.
    What it says now
    Resolved as a module boundary rather than a stale field, and the platforms, interfaces and jobs fields were left unchanged. The vendor family page lists AccessibilityNow Publisher, Gateway, Designer, Validator, Remediate, Dashboard, DocMD, Language Translation Services and AccessibilityNow.com as separate modules and uses SaaS, cloud, server, browser and web-based language, while the Remediate page says AccessibilityNow Remediate is Crawford Technologies ad hoc tagging tool that runs in a Windows desktop environment and is a graphical user interface tool, straplined as a Tagging GUI Driven Tool for Non-Transactional Documents. Because this record covers the family, whose delivery the note already describes as cloud-first with on-premises options, a Windows desktop interface on one module does not contradict web and self_hosted platforms. The deployment note now records that boundary and no longer says no further detail was read. Sources were re-retrieved by direct request on 2026-10-16.

  105. 7a053ac

    The Adobe Acrobat record, in the primary edit job

    What it said
    The edit job rested on the record own prose, a prose-only candidate signal, so it could be read as arbitrary editing, OCR, conversion or page manipulation.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing existing PDF content in Acrobat desktop by adding, changing, replacing or deleting text, formatting text and adjusting text boxes, editing numbered or bulleted lists, and adding, modifying or deleting images and objects, as documented by the Acrobat desktop help task pages, from https://helpx.adobe.com/acrobat/desktop/edit-documents/edit-text-in-pdfs/modify-text.html read on 2026-10-16. That page is part of the official Adobe Acrobat Desktop Help, whose Edit documents chapter lists task pages titled Add new text, Change replace or delete text, Format text, Adjust text boxes, Set default fonts, Edit numbered or bulleted lists and Check spellings, plus Add images, Modify images, Edit images with external applications, Add objects and Delete images. Retrieval was limited: direct requests to helpx.adobe.com return 403 and the fetch tool truncates every page after the navigation, so the evidence is Adobe own help structure and task titles rather than body prose, and that limitation is recorded rather than hidden. The scope claims no arbitrary editing, no OCR, no conversion, no page manipulation, no annotation editing, no graphical interface breadth and no edition beyond Acrobat desktop. Other jobs and fields are unchanged.

  106. 7a053ac

    The Adobe Acrobat record, in the primary edit job, evidence corrected

    What it said
    The previous pass scoped the edit job from Adobe help navigation and task page titles after direct requests to helpx.adobe.com returned 403 and the fetch tool truncated those pages after the navigation. Title level navigation is a candidate signal in this audit, not capability evidence, so that scope stood on a weaker basis than the rest.
    What it says now
    The scope was restated on body evidence from an official Adobe page that does render, https://www.adobe.com/acrobat.html, read on 2026-10-16, whose body says Edit PDFs in a few clicks, updating text and images, rearranging content, organizing pages and more, and says Acrobat Standard allows you to edit and organize PDFs while Acrobat desktop software provides tools to edit, convert, protect and sign documents. The scope now reads editing PDF content by updating text and images and rearranging content, in Adobe Acrobat desktop, included in the Standard and Pro plans, and deliberately omits organizing pages and the scanned document wording that the page also carries. The job was not removed and no other Adobe field was changed.

  107. 7a053ac

    The Adobe Acrobat record, in the compress job

    What it said
    The compress job carried no scope, so it could be read as lossless optimisation, a size guarantee, batch compression or an algorithm claim.
    What it says now
    The job keeps compress and gains the scope compressing PDFs by selecting a High, Medium or Low compression level, which Acrobat applies by reducing image resolution, flattening interactive layers and removing unused metadata, in the Acrobat online tool and desktop app, where unlimited compression requires a paid plan, from https://www.adobe.com/acrobat/online/compress-pdf.html read on 2026-10-16. That page is Adobe own compression page on the route that renders; its title says Compress PDFs for free and Reduce PDF file size online, its how to section says to select a file to make smaller and choose the compression level after which Acrobat automatically reduces the file size, and its explanation of the mechanism names reducing image resolution, flattening interactive layers and removing unused metadata with High, Medium and Low levels and a trade off between quality and size. Its FAQ adds that compression removes redundant data and optimises images, preserves hyperlinks, form fields and digital signatures, compresses locally in the desktop app and on servers online, and does not support password protected files unless the password is entered first. The scope claims no lossless behaviour, no size or quality guarantee, no percentage figure, no batch behaviour and no algorithm detail, and its edition boundary is only the free versus paid distinction the page states. The corrected edit scope and every other field are unchanged.

  108. 7a053ac

    The Adobe Acrobat record, in the convert job

    What it said
    A ledger entry committed this pass claimed that the convert job gained a scope in that pass. That claim was wrong: the job already carried a scope before the pass began, and the pass changed no content at all.
    What it says now
    Corrected. The convert job already carried the scope Word, PowerPoint, Excel, Publisher, text, image and other supported files to PDF in Reader; export is a separate feature, settled in an earlier pass from https://helpx.adobe.com/reader/desktop/create-pdf.html, and this pass did not alter it: the scope update was refused because a scope was present, so no content changed. The two www.adobe.com pages read this pass, https://www.adobe.com/acrobat.html and https://www.adobe.com/acrobat/online/compress-pdf.html, both read with body text on 2026-10-16 on the www.adobe.com route that renders. The product page says under product capabilities Quickly create and convert PDFs, turning Microsoft files, images and more into high fidelity PDFs or converting PDFs to other formats, its plan description says Acrobat Standard includes converting documents to and from PDF, and its subscription section says Acrobat desktop software provides tools to edit, convert, protect and sign documents. The compress page FAQ says that after compressing a PDF you can convert PDFs to other file formats, merge and split files and fill and sign documents. do corroborate conversion: the product page says under product capabilities Quickly create and convert PDFs, turning Microsoft files, images and more into PDFs or converting PDFs to other formats, and Acrobat Standard includes converting documents to and from PDF. A named tool page attempt, pdf to word, redirected to the web application and was not used. The existing Reader scope stands unchanged, and the edit and compress scopes and every other field are unchanged.

  109. 7a053ac

    The Adobe Acrobat record, in the ocr job

    What it said
    The ocr job carried no scope and the inventory entry read unresolved with a prose only signal, so the label could be read as language coverage, an accuracy claim, batch behaviour or PDF/A output.
    What it says now
    The job keeps ocr and gains the scope recognising text in scanned documents to turn them into editable, searchable PDFs, documented for Acrobat Pro and for the Acrobat Scan mobile app which recognises text automatically, from https://www.adobe.com/acrobat.html read with body text on 2026-10-16. That page says under product capabilities Go from print to digital, easily turning scanned paper docs into fully searchable and editable PDFs, says Acrobat Pro includes turning scanned documents into editable searchable PDFs, says the Action Wizard in Acrobat Pro can run sequences of tasks like OCR redaction watermarking and file conversion, and describes the Acrobat Scan mobile app as recognising text automatically, in parentheses OCR. The plan boundary is stated because the page states it. The scope claims no language coverage, no accuracy figure, no batch behaviour, no PDF/A output and no local versus cloud processing statement. The edit, compress and convert scopes and every other field are unchanged.

  110. 7a053ac

    The Adobe Acrobat record, in the accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as a conformance guarantee, a certification, automatic remediation of every issue or breadth beyond PDF/UA and WCAG.
    What it says now
    The job keeps accessibility and gains the scope checking PDF accessibility against PDF/UA and WCAG with the Prepare for accessibility and Check for accessibility tools, including an accessibility report, reading options and the Reading Order tool, in Acrobat Pro, from https://helpx.adobe.com/acrobat/using/create-verify-pdf-accessibility.html read on 2026-10-16. That page says Acrobat can make PDFs meet common accessibility standards such as the latest Web Content Accessibility Guidelines and PDF/UA, described as Universal Access or ISO 14289, and lists the accessibility tools including Prepare for accessibility, which automates many tasks and checks accessibility with instructions for manual fixes, Check for accessibility, which verifies conformance to PDF/UA and WCAG 2.0, Open accessibility report, Reading options, the Reading Order tool and Save as accessible text. It adds that the check runs on all PDFs except dynamic XFA forms and portfolios, looks for scanned text, form fields, tables and images, and reports Passed, Skipped By User, Needs Manual Check or Failed, with the caveat that it does not distinguish essential from nonessential content. The scope claims no conformance guarantee, no certification, no automatic fixing of every issue, no ADA breadth and no edition breadth beyond Acrobat Pro and the versions named. The other Adobe scopes and fields are unchanged. The page was reached on the legacy acrobat/using path, which renders body text where the modern acrobat/desktop paths did not.

  111. 7a053ac

    The Adobe Acrobat record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as certificate encryption, DRM, an encryption strength claim, redaction or a compliance guarantee.
    What it says now
    The job keeps protect and gains the scope password-protecting PDFs so that only people with the password can open them, with permission controls that restrict actions such as editing, copying and printing, in the Acrobat online tools and in Acrobat Pro, from https://www.adobe.com/uk/acrobat/online/password-protect-pdf.html read on 2026-10-16. That page says a PDF can be password protected in four steps by uploading it, entering and confirming a password, clicking Set password and downloading the result, that clicking Add a Password encrypts the PDF and restricts access to authorised people, and in its FAQ that password protection restricts access while encryption protects the contents, that the tool supports password protection and permission controls, that it limits actions like editing or printing, that Acrobat Pro can remove a password with the Unlock tool, and that forgotten passwords cannot be recovered. The scope claims no certificate encryption, no DRM, no algorithm or strength, no redaction, no compliance guarantee, no size limit and no edition breadth beyond the online tools and Acrobat Pro. The other Adobe scopes and fields are unchanged. A legacy acrobat/using passwords URL now redirects to the truncating modern path, so this evidence comes from the www.adobe.com tool page.

  112. 7a053ac

    The Adobe Acrobat record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as covering PAdES, certificate profiles, trust validation, timestamping, algorithms or legal validity, and it did not distinguish electronic signature workflows from certificate-based digital signatures.
    What it says now
    The job keeps sign and gains the scope signing PDFs with an electronic signature by typing, drawing or uploading an image of a signature, and with certificate-based digital signatures, in the free Acrobat online sign tool and the Acrobat Reader mobile app, where requesting signatures from others and audit trails may require a paid plan, from https://www.adobe.com/acrobat/online/sign-pdf.html read on 2026-10-16. That page describes filling and signing a PDF with an e-signature through the Sign panel, says an electronic signature can be created for free and placed anywhere with a typed, drawn or uploaded image, and its FAQ records the paid plan boundary for requesting signatures and audit trails, describes Adobe Acrobat Sign as a separate dedicated service for higher volume workflows, complex routing, bulk send and enterprise authentication, and notes that the Acrobat Reader mobile app includes signing features. The page explicitly distinguishes an electronic signature from a digital signature that uses a certificate-based digital ID, and states that Acrobat supports both, which is why the scope names both types and identifies the tools. It claims no PAdES, no certificate profiles, no trust validation feature, no timestamping, no algorithms and no legal validity, and no edition breadth beyond the free online sign tool and the Reader mobile app, with the paid boundary recorded as the page states it. The other Adobe scopes and fields are unchanged.

  113. 7a053ac

    The ABBYY FineReader PDF record, in the sign job, retry result

    What it said
    The sign job was left unresolved after the Windows guide showed a Signing PDF documents section but its digital signature article body did not load. This entry records what the retry found rather than leaving the blocker undated. It cites the last commit that touched this record, because the retry changed no content.
    What it says now
    Still unresolved, and the blocker is specific. The Windows FineReader 14 user guide exposes a Signing PDF documents section whose pages are Digital signature, Text signature and Picture signature. The sibling text signature page was retried on the finereader/14/user_guide route and returned terms of use and navigation only, exactly as the digital signature page did, so the Windows guide pages behave differently from the Mac guide pages on the same host, which do render bodies. ABBYY support centre searches for signature and PDF returned no usable body text and one support section URL answered 403. A static user guide PDF asset was considered but is not readable as text with the retrievers available here. The job is not removed and stays unresolved, because a section existing is navigation rather than capability evidence and no body sentence about signing was reachable. The settled ABBYY scopes for ocr, convert, protect and edit are untouched.

  114. 7a053ac

    The Adobe PDF Accessibility Auto-Tag API record, in the primary accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as full PDF/UA or WCAG conformance, certification, manual remediation or an accuracy guarantee.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope automatically tagging PDFs to add document structure, replacing any existing tags, with an optional report on the added tags, through the REST-powered PDF Accessibility Auto-Tag API in the Adobe PDF Services API, where the tagged output is not guaranteed to comply with WCAG or PDF/UA without further remediation, from https://developer.adobe.com/document-services/docs/overview/pdf-accessibility-auto-tag-api/ read on 2026-10-16. That page says tags define the reading order and identify headings, paragraphs, sections and tables, that the API is part of the Adobe PDF Services API as a REST-powered SDK with Sensei AI, usable for individual PDFs and at scale, that automatic tagging is a good first step that outputs the tagged PDF with an optional report and replaces existing tags, and that further work such as alt-text, reviewing complex tables, checking the title and reading order and resolving report errors may still be needed. Its options are GenerateReport and ShiftHeadings. Crucially the page states that the output is a tagged PDF but is not guaranteed to comply with WCAG or PDF/UA without further remediation, and the scope carries that limit in the same sentence. It claims no certification, no guaranteed conformance, no manual remediation, no accuracy figure, no ADA breadth and no platform breadth beyond the API and SDK named. Other fields are unchanged.

  115. 7a053ac

    The Adobe PDF Extract API record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as OCR, arbitrary object extraction, an accuracy or layout fidelity guarantee, batch behaviour or output types the vendor does not name.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting text in contextual blocks, complex tables and figures from native or scanned PDFs as structured JSON or Markdown through the cloud-based PDF Extract API, called via the PDFServices SDK for Node.js, Python, .NET and Java, with optional CSV, XLSX and PNG outputs and a free tier of 500 document transactions per month, from https://developer.adobe.com/document-services/docs/overview/pdf-extract-api/ read on 2026-10-16. That page says the suite is included with the PDF Services API and is a cloud based web service using Sensei AI to extract content and structural information from native or scanned PDFs, offering structured JSON and Markdown outputs, both of which extract text in contextual blocks such as paragraphs and headings with font and styling information, complex tables parsed per cell with automatic detection of cells spanning rows or columns, and figures or images as PNG or base64. It adds that Extract JSON can optionally deliver table data as CSV and XLSX and tables as PNG images, that the JSON captures natural reading order and per page layout, that the API embeds via the PDFServices SDK for Node.js, Python, .NET and Java, and that a free tier includes 500 document transactions per month. The scope claims no OCR, no accuracy figure, no layout fidelity beyond the page wording, no batch behaviour, no arbitrary object extraction and no unnamed output types. Other fields are unchanged.

  116. 7a053ac

    The Adobe PDF Extract API record, in the ocr job

    What it said
    The ocr job carried no scope. The risk was inferring OCR from the word scanned in the overview, which would have been an inference rather than evidence.
    What it says now
    The job keeps ocr and gains the scope extracting text from scanned PDFs through the cloud PDF Extract API, whose own documentation discusses OCR and scan quality as dependent on input clarity such as skewed pages, shadowing, obscured or overlapping fonts and resolution below 200 DPI, from https://developer.adobe.com/document-services/docs/overview/pdf-extract-api/ and its extract how-to page, both read on 2026-10-16. The overview alone describes extraction from native or scanned PDFs, which would not justify an OCR claim. The how-to page carries a section headed OCR and Scan quality stating that the quality of text extracted from scanned files depends on the clarity of content in the input file and naming skew, shadowing, obscured or overlapping fonts and resolution below 200 DPI as limiting conditions, which is Adobe associating the service scan handling with OCR rather than merely using the word scanned. The scope claims no accuracy guarantee and keeps the named limiting conditions, no language coverage despite the page discussing English optimisation, no searchable output claim, no batch behaviour and no API boundary beyond the cloud PDF Extract API. The extract scope and other fields are unchanged. Note that the Extract API pages contain exactly one OCR mention, in that section heading and sentence.

  117. 7a053ac

    The Adobe Scan record, in the primary ocr job

    What it said
    The ocr job carried no scope, so it could be read as language coverage, an accuracy claim, a searchable quality guarantee, PDF/A output or batch behaviour, and it did not state the mobile platforms.
    What it says now
    The job keeps ocr, remains primary, and gains the scope converting scanned documents into editable, searchable PDF files using OCR, optical character recognition, in the Adobe Scan app on iOS 15 or later and Android 8.0 or later, with scans stored in Adobe Document Cloud, from https://www.adobe.com/acrobat/mobile/scanner-app.html read on 2026-10-16. That page says under AI-enhanced scans that Adobe Scan uses AI to correct image perspective, sharpen handwritten or printed text and remove glares and shadows, and that with OCR you can convert scanned documents into editable, searchable PDF files instantly and easily extract text. It also describes capture modes and automatic boundary detection, and its FAQ says Adobe Scan automatically detects edges, cleans backgrounds and enhances text clarity, integrating with Acrobat so scans are instantly editable and searchable. The FAQ lists compatibility as iOS 15 or later and Android 8.0 or later and says scans are stored in Adobe Document Cloud storage. The scope claims no language coverage, no accuracy figure, no searchable quality guarantee beyond the vendor phrase for the output, no PDF/A output and no batch behaviour. The plan boundary is not folded in because the OCR capability is described for the free app. Other fields are unchanged.

  118. 7a053ac

    The Adobe Scan record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as arbitrary format conversion, round trip fidelity, batch behaviour or PDF/A output.
    What it says now
    The job keeps convert and gains the scope converting captures of documents, forms, business cards and whiteboards into PDFs in Adobe Scan, and exporting scanned files from Adobe Scan as PDF, JPEG or Word, from https://www.adobe.com/acrobat/mobile/scanner-app.html read with body text on 2026-10-16. That page says that with the Adobe Scan app you can easily capture and convert documents, forms, business cards and whiteboards into high quality PDFs, and its FAQ answers how to export scanned files by opening a scanned file in Adobe Scan, tapping Share or Open in Acrobat, choosing the export format from PDF, JPEG or Word, and saving to local storage, an Adobe account or a connected cloud service. It also mentions saving files as JPEGs. The scope covers the capture to PDF direction and the export direction the page states, and nothing else. It claims no arbitrary or other format conversion, no round trip fidelity, no batch behaviour, no PDF/A output, no platform breadth and no paid plan requirement, since both directions are described for the free app while the paid trial is attached to other Acrobat Pro capabilities. The ocr scope and other fields are unchanged.

  119. 7a053ac

    The Allyant PDF Validator record, in the primary accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as automatic remediation, certification, ADA breadth or a cross platform checker.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope validating PDF accessibility as an Adobe Acrobat plugin that reports issues and guides manual tests, covering tagging and reading order, alternative text, headings, lists, links, form tags, tooltips, tables, metadata, font embedding and Unicode mapping, with the vendor naming Section 508, WCAG, PDF/UA and HHS guidelines, on Windows with Acrobat Standard or Pro X or higher, from https://allyant.com/commonlook-accessibility-suite/cl-pdf-validator/ read on 2026-10-16. The page describes a free Adobe Acrobat plugin for testing and verifying PDFs for accessibility that identifies and highlights accessibility concerns, and says it does not stop at automated checks but guides the manual tests necessary for achieving complete accessibility compliance. Its features are verification tasks covering metadata, title visibility, font embedding, Unicode mapping, the tags tree and logical reading order, tag nesting, heading tags, tag based content highlighting, tag properties including alternative text, untagged content, list and link tags, link alt text and contents, broken links, form tags and tooltips, and tables with their tags and summaries. It names Section 508, WCAG, PDF/UA and HHS guidelines and requires Windows 8 or higher with Acrobat Standard or Pro version X or higher. The scope claims no remediation or automatic fixing, since the page itself stops at checks and manual guidance; no certification, although the page mentions a certification report, because that is a vendor label rather than a certified outcome and the instruction excludes certification; no ADA breadth beyond the standards named; no accuracy figure; and no platform breadth beyond Windows with Acrobat. Other fields are unchanged.

  120. 7a053ac

    The Amazon Textract record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as OCR accuracy, arbitrary document and object coverage, batch behaviour or language breadth.
    What it says now
    The job keeps extract, remains primary, and gains the scope detecting typed and handwritten text and extracting text, forms and tables, and extracting specified information with queries, through the Amazon Textract Document Analysis and Analyze Document APIs on image and PDF files, from https://docs.aws.amazon.com/textract/latest/dg/what-is.html read on 2026-10-16. The page says the service adds document text detection and analysis, that typed and handwritten text can be detected in documents including financial reports, medical records and tax forms, that text, forms and tables can be extracted from structured documents with the Document Analysis API, that information can be specified and extracted with the Queries feature in the Analyze Document API, and that the service analyses image files and PDF files. It adds that text can be extracted as words and lines and grouped by table cells when table analysis is enabled, and that synchronous processing handles single page documents while asynchronous operations extend to multipage documents. The scope claims no OCR accuracy figure, no arbitrary document or object coverage, no batch behaviour and no language breadth. The specialised AnalyzeExpense, AnalyzeID and Analyze Lending workflows and Custom Queries are documented on the page but are deliberately not folded into the scope, which states the narrowest set of outputs; they are recorded here instead. The input boundary is the page wording, image and PDF files. Other fields are unchanged.

  121. 7a053ac

    The Amazon Textract record, in the secondary ocr job, and its tagline

    What it said
    The ocr job carried no scope. The tagline also needed checking: the previous extraction pass found that the introduction page does not mention signatures, so the tagline clause about signatures was unverified rather than wrong.
    What it says now
    Two outcomes. First, ocr keeps the job and gains the scope detecting typed and handwritten text in images and PDFs through Amazon Textract text detection, with synchronous processing for single page documents and asynchronous operations for multipage documents, from https://docs.aws.amazon.com/textract/latest/dg/what-is.html read on 2026-10-16, which says typed and handwritten text can be detected in a variety of documents, that the service analyses image files and PDF files, and that synchronous processing handles single page documents while asynchronous operations extend to multipage documents. Second, the tagline is verified and unchanged: https://docs.aws.amazon.com/textract/latest/APIReference/API_AnalyzeDocument.html read on 2026-10-16 states under the types of information returned that Signatures, a SIGNATURE Block object, contains the location information of a signature in a document and can be given a Key-Value pairing or detected in a table cell, and lists the FeatureTypes valid values as TABLES, FORMS, QUERIES, SIGNATURES and LAYOUT with the instruction to add SIGNATURES to return the locations of detected signatures. The tagline wording about signatures is therefore directly supported, and its text, forms and tables clauses are supported by the introduction page while uploaded documents matches the Document parameter accepting bytes or an S3 object. The same API page also documents input formats as JPEG, PNG, PDF or TIFF and notes AnalyzeDocument is synchronous with StartDocumentAnalysis for asynchronous use. The scope claims no accuracy figure, no language breadth, no every document type, no handwriting quality, no batch behaviour and no analysis features beyond detection; the LAYOUT feature is recorded here but not folded into a scope. The settled extract scope and other fields are unchanged.

  122. 7a053ac

    The Apache PDFBox record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as batch merging, a page order guarantee, losslessness, preservation of bookmarks or forms, or a graphical feature.
    What it says now
    The job keeps merge and gains the scope merging a list of PDF documents into a new document with the PDFMerger command line tool in PDFBox, taking two or more source PDF files and a target PDF file, from https://pdfbox.apache.org/2.0/commandline.html read on 2026-10-16. The page says under Merge documents that the application takes a list of pdf documents and merges them, saving the result in a new document, with usage java -jar pdfbox-app-2.y.z.jar PDFMerger source PDF files, two to n, and a target PDF file, and it presents these as command line utilities available as standard Java applications. The scope claims no batch behaviour, no page order guarantee, no losslessness, no preservation of bookmarks or forms and no graphical feature, none of which the page states. The input and output boundaries are the page wording. The settled edit, convert, extract and compress scopes, the tagline, the summary and every other field are unchanged, and no prose claim needed correcting: the record prose that mentions merging is consistent with the tool description rather than contradicting it.

  123. 7a053ac

    The Apache PDFBox record, in the primary protect job

    What it said
    The protect job carried no scope, so it could be read as generic DRM, as naming a modern encryption algorithm, as promising unbreakability, or as covering security modes the tool does not document.
    What it says now
    The job keeps protect, remains primary, and gains the scope encrypting PDFs with the PDFBox Encrypt command line tool using owner and user passwords or an X.509 certificate file, with permission settings for assembling, extracting content, filling in forms, modifying, annotating, printing and printing degraded, and key lengths of 40, 128 or 256 bits, and decrypting PDFs with the Decrypt tool using the owner password or a certificate keystore, from https://pdfbox.apache.org/2.0/commandline.html read on 2026-10-16. That page documents Encrypt with -O owner password, -U user password, -certFile as a path to an X.509 cert file, the permission flags canAssemble, canExtractContent, canExtractForAccessibility, canFillInForm, canModify, canModifyAnnotations, canPrint and canPrintDegraded each defaulting to true, and -keyLength with the documented values 40, 128 or 256 and the note that 128 and above require the unlimited strength policy files. It documents Decrypt with -password and with -keyStore and -alias for a certificate keystore, noting that the owner password is required and that a keystore is needed only when a certificate was used. The scope claims no generic DRM, no algorithm name because the page documents key lengths rather than algorithms, no modern default, no unbreakability and no security mode beyond encryption and decryption. The source is the 2.0 command line documentation, which is the version boundary. The merge scope settled earlier in the session, the other scopes, the tagline, the summary and every field are unchanged.

  124. 7a053ac

    The Apache PDFBox record, in the extract job

    What it said
    The extract job carried no scope, so it could be read as OCR, as guaranteeing layout fidelity, as covering arbitrary file formats, as extracting tables or form data, or as a batch operation.
    What it says now
    The job keeps extract and gains the scope extracting all text from a PDF with the ExtractText command line tool in PDFBox, writing raw text, HTML or Markdown, with encoding, page range, console output and sort options, from https://pdfbox.apache.org/2.0/commandline.html read on 2026-10-16. The page says the Extract text application extracts all text from the given PDF, with usage ExtractText plus options, an input file and an optional text file. Its documented options are -alwaysNext, -password, -encoding with examples ISO-8859-1, UTF-8 and UTF-16BE, -console to send text to the console, -html for HTML output, -md for Markdown output since 2.0.33, -sort to sort text before writing, -ignoreBeads, -debug, -rotationMagic for rotated or skewed text, and -startPage and -endPage as one based bounds. The scope claims no OCR, no layout fidelity, no arbitrary file formats, no table or form extraction, no batch behaviour and no preservation guarantee. The rotationMagic option is recorded here but not carried into the scope, and the page separate form data capability is not part of it. The version boundary is the 2.0 command line documentation. The merge and protect scopes and every other field are unchanged. The record summary wording about extracting text is consistent with this tool and needed no correction.

  125. 7a053ac

    The Apache PDFBox record, in the view job

    What it said
    The view job carried no scope. The risk was mapping it to PDFDebugger, which inspects internal structure rather than rendering pages, and the record also needed a rendering surface to be identified.
    What it says now
    The job keeps view and gains the scope rendering each page of a PDF to an image file with the PDFToImage command line tool in PDFBox, writing JPG or PNG at a selectable DPI and colour depth, with page range and crop box options, from https://pdfbox.apache.org/2.0/commandline.html read on 2026-10-16. The page documents PDFDebugger as taking an existing PDF and allowing you to analyse and inspect the internal structure, described as the replacement for the PDFReader removed in 2.0.0, which is inspection rather than viewing and is therefore not the basis for this scope. The separate PDFToImage application creates an image for every page in the PDF document, with an image type limited to jpg or png, an output prefix, one based start and end page bounds, a single page selector, DPI resolution, colour depth values bilevel, gray, rgb and rgba, a crop box and optional timing output. The scope claims no graphical interface and no on screen viewing because the tool writes image files, no rendering format beyond JPG and PNG which are the only two listed, no visual fidelity guarantee and no batch behaviour. The version boundary is the 2.0 command line documentation. The settled merge, protect, extract, convert, edit and compress scopes and every other field are unchanged.

  126. 7a053ac

    The Apache PDFBox record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as PAdES support, as timestamping, trust validation or legal validity, or as an algorithm claim.
    What it says now
    The job keeps sign and gains the scope adding a digital signature to a PDF with the PDFBox signature API, using a PDSignature dictionary with the Adobe PPKLite filter and the adbe.pkcs7.detached subfilter, a keystore, and an incremental save, as shown in the project own CreateSignature and CreateVisibleSignature examples, from the Apache PDFBox repository example sources retrieved on 2026-10-16 at https://raw.githubusercontent.com/apache/pdfbox/2.0.32/examples/src/main/java/org/apache/pdfbox/examples/signature/CreateSignature.java and the trunk copy of the same file plus CreateVisibleSignature.java. Retrieval route: the official Command-Line Tools page has no signing tool, the Examples page is a stub that defers to the examples directory, and the javadoc proxy returned only its frame, so the repository example source was the reachable primary source. The source builds a PDSignature, calls setFilter with FILTER_ADOBE_PPKLITE and setSubFilter with SUBFILTER_ADBE_PKCS7_DETACHED, sets name, location and reason, uses a KeyStore, and references SignatureInterface, SignatureOptions and saveIncremental, with a guard against modifying a document whose DocMDP transform parameters forbid it. CreateVisibleSignature demonstrates a visible signature. The scope claims no PAdES because the documented subfilter is adbe.pkcs7.detached, no timestamping although the sources reference a timestamp authority, no trust validation, no legal validity, no algorithm or key list, no graphical interface and no batch behaviour. The version boundary is the 2.0.32 and trunk example code. The record summary wording about digitally signing documents is supported by this source and needed no correction. The seven settled scopes and every other field are unchanged.

  127. 7a053ac

    The Apache Tika record, in the primary extract job

    What it said
    The extract job carried no scope, so the breadth in the record prose could be read as arbitrary format support, layout fidelity, table extraction, batch behaviour or OCR.
    What it says now
    The job keeps extract, remains primary, and gains the scope detecting and extracting metadata and text from over a thousand different file types, including PPT, XLS and PDF, parsed through a single interface, in the Apache Tika toolkit, from https://tika.apache.org/ read on 2026-10-16. The page describes Apache Tika as a content analysis toolkit and says the toolkit detects and extracts metadata and text from over a thousand different file types, gives PPT, XLS and PDF as examples, and says all of those file types can be parsed through a single interface, making Tika useful for search engine indexing, content analysis and translation. It links documentation for the 4.1.x line and the supported 3.3.x maintenance line, and its release notes record that 4.0.0 switched the default output to Markdown and added vision language model parsers and in-process Tesseract OCR. The scope uses the vendor own figure rather than claiming arbitrary format support, and claims no layout fidelity, no table extraction, no batch behaviour and no OCR. The OCR and language model parsers are recorded here but deliberately kept out of this scope, and the separate secondary ocr job is untouched. The record summary and tagline were compared with this evidence and are consistent, so they are unchanged.

  128. 7a053ac

    The Apple Preview record, in the primary view job

    What it said
    The view job carried no scope, so it could be read as covering editing, markup, rendering export, OCR or arbitrary file formats.
    What it says now
    The job keeps view, remains primary, and gains the scope opening PDFs and images in the Preview app on Mac, by double-clicking a file or using File Open, changing how documents are shown in the Preview window and getting information about the files, from https://support.apple.com/en-nz/guide/preview/prvw11470/11.0/mac/11.0 read on 2026-10-16. Retrieval note: the fetch tool reported an unsupported content type for Apple guide URLs and direct requests to three guide shell URLs returned the same navigation only, so the article was located by search and then retrieved by direct request. The page says you can open PDFs and images in Preview, change how documents are shown in the Preview window and get information about the files; under Open PDFs and images it says a PDF or image can be double-clicked to open by default in Preview, or Preview opened and files selected to view, or File then Open used in the Preview app on Mac, and mentions File then Open Recent and File then Open With then Preview. The guide table of contents also lists separate pages for finding text, slide shows, animated GIF frames, locking, printing and, in other sections, markup and image editing. The scope claims no editing, markup or annotation, no rendering or export, no OCR, no browser behaviour and no every file format, and states the Mac platform boundary. The tagline and summary are consistent with this evidence and unchanged.

  129. 7a053ac

    The Apple Preview record, in the merge job, retry result

    What it said
    The merge job has no scope. The guide table of contents lists a Combine PDFs entry, which is navigation rather than evidence, and the article body had not been retrieved. This entry records the retry so the blocker is not left undated.
    What it says now
    Still unresolved, and the attempted route produced a warning rather than evidence. The user guide content is client rendered: the fetch tool reports an unsupported content type for Apple guide URLs, direct requests to the guide shell return the same navigation only, and the table of contents page renders its links client side so a direct request returns only twenty links and no Combine entry. A search for the official article surfaced only third party pages, which are not acceptable evidence here. Probing a likely slug from the observed pattern, combine-pdfs-prvw2015, returned HTTP 200 but the body was the article Resize, rotate or flip an image in Preview on Mac, so Apple serves a default article for an unknown slug and a 200 status from a guessed URL is not proof that the requested article exists. Trusting that response together with the table of contents entry would have scoped merging from an article about resizing images, so the job is left unresolved rather than scoped from a title or a soft 404. It is not removed, because nothing contradicts it and the guide does list a Combine PDFs entry. The settled view, sign, edit, protect, convert and compress scopes are untouched.

  130. 7a053ac

    The Apryse record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as reflow, as editing arbitrary formats, as lossless, as batch behaviour or as client side editing across the whole product family.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing text and images directly within PDF files using WYSIWYG PDF content editing in the Apryse WebViewer JavaScript library from version 10.3, changing, replacing, deleting, moving, resizing and reformatting existing content and adding new paragraphs and images, as a separate PDF Editing add-on that works with PDF files only, from https://docs.apryse.com/web/edit-page-content/text-edit.md read on 2026-10-16. Retrieval note: the Apryse docs are published with GitBook and render client side, so the plain page URL returned only the shell, while appending .md to a documentation page URL returns the full Markdown text, which is the route used here. The page says that from WebViewer 10.3 text and images can be edited directly within PDF files with WYSIWYG content editing, in the UI through an Edit Text toolbar group, dashed content boxes and a Text Styles panel, supporting move, resize, delete, editing existing text with font, size, colour and alignment, adding paragraphs and adding images, and programmatically through ContentEditManager start and end content edit mode as of 12.2.0, AddParagraphTool, AddImageContentTool, getContent, updateDocumentContent, getTextAttributes, setTextAttributes, the ContentEditHistoryManager undo and redo, and link annotation detection and deletion. Vendor boundaries: an info note says PDF Editing is a separate add-on not included in a standard WebViewer license, and a warning says content editing only works with PDF files, other types having to be loaded as PDF. The scope therefore names WebViewer, states the add-on and PDF-only boundaries, and claims no reflow, no arbitrary formats, no losslessness, no batch behaviour and no client side availability across the family. The tagline and summary are consistent with the evidence and unchanged.

  131. 7a053ac

    The Apryse record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as certificate-based encryption, as enforced permissions, as DRM, as unbreakable, or as an algorithm claim beyond what is documented.
    What it says now
    The job keeps protect and gains the scope password-based PDF encryption with user passwords at AES-128 or AES-256, and setting permissions that the Standard Security Handler records but does not enforce, requiring the Full API and Security package for production use, in Apryse WebViewer, from the Apryse documentation retrieved on 2026-10-16: the security page at https://docs.apryse.com/web/get-started/guides/features/security.md, which is only a link to the custom security handler page described in the index as protecting files for DRM use cases and as a built-in SDK feature for encrypting and decrypting documents, and the documentation own query endpoint on the same URL, which answers that WebViewer supports password-based PDF encryption, permissions and certificate-based digital signatures and certification. On encryption it names user passwords and AES-128 or AES-256 and says production use requires the Full API and Security package. On permissions it says permissions can be set but the Standard Security Handler only records them and does not enforce them, so the application must honour them. On certificates it says support covers signing and integrity rather than certificate-based encryption and that encryption to recipient certificates is not confirmed. The scope therefore excludes certificate-based encryption, DRM, unbreakability, algorithms beyond the AES key lengths named, and compliance claims, and it keeps redaction and signing separate. Retrieval note: the docs are GitBook hosted, the llms.txt index advertised on each page was used to locate the security section, and the ask endpoint is a mechanism the documentation itself documents. Other fields are unchanged.

  132. 7a053ac

    The Apryse record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as PAdES, as a claim of legal validity, as trust validation, as certificate-based encryption, or as covering timestamps and long term validation as part of signing.
    What it says now
    The job keeps sign and gains the scope digitally signing and certifying PDFs in Apryse WebViewer, using PKI signing with PFX digital certificates or self-signed certificates, importing and exporting signature fields and data through XFDF or FDF, and signing with images, ink annotations or custom appearances, with document timestamping and long term validation documented as separate features and the Digital Signature package required for production use, from https://docs.apryse.com/web/digital-signature/signature.md read on 2026-10-16. That page, which is the exact path the security answer named, presents the library as providing authentication, integrity and non-repudiation; says the Digital Signature package is required for production while trial keys have unlimited access; lists benefits including XFDF and FDF signature field import and export, built-in PKI signing with PFX digital certificates, custom signature handlers, and signing with images, ink annotations or custom appearances; distinguishes an e-signature annotation from a digital signature using a cryptographic algorithm; explains that a CA is not required and a self-signed certificate can be used, with Apryse not providing CA services; documents certificate expiration and revocation plus Document Timestamping and Long Term Validation as two separate features; and states that WebViewer can digitally sign and certify a PDF, with separate guides for signing, certifying, verifying, custom signing and embedded timestamping. The scope claims no PAdES, no legal validity, no trust validation feature, no certificate-based encryption, no algorithm list and no GUI or batch behaviour, and names timestamping and LTV only as separate features. Retrieval note: the per-page llms.txt index and the .md route are the working ways into these GitBook docs. Other fields are unchanged.

  133. 7a053ac

    The Aspose.PDF record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as reflow, as editing arbitrary formats, as lossless, as batch behaviour or as cross language support.
    What it says now
    The job keeps edit, remains primary, and gains the scope replacing existing text in a PDF with Aspose.PDF for .NET, using TextFragmentAbsorber to locate text fragments across pages and setting the replacement text with font, size and colour before saving, on .NET Core or .NET Framework 4.7.2 and above, where the documentation warns that replacement text exceeding page boundaries can lead to layout issues, from https://tutorials.aspose.com/pdf/net/text-operations/aspose-pdf-net-text-replacement-guide/ read on 2026-10-16. That official Aspose tutorial describes installing Aspose.PDF for .NET, opening and manipulating PDFs programmatically, replacing specific text in C# and customising appearance afterwards. It requires a system with .NET, preferably .NET Core or .NET Framework 4.7.2 and above. It loads a document with the Document class, uses a TextFragmentAbsorber with Pages.Accept to find text, iterates TextFragments setting Text and TextState font, font size, foreground and background colour, and saves the result. It warns in troubleshooting that replacement text must not exceed page boundaries because that may lead to layout issues, and its FAQ says the absorber searches all pages by default. The scope therefore claims no reflow, carries the vendor layout warning, and excludes arbitrary formats, losslessness, batch behaviour and cross language support, since the source names the .NET runtime. Annotations, forms, redaction and conversion are separate areas and are not folded in. Other fields are unchanged. Retrieval note: the tutorial lives on the Aspose tutorials subdomain and links the reference documentation at reference.aspose.com.

  134. 7a053ac

    The Aspose.PDF record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as arbitrary format conversion, as round trip fidelity, as lossless, as batch behaviour, as PDF/A or PDF/UA conformance, or as cross language support.
    What it says now
    The job keeps convert and gains the scope converting PDF documents to Microsoft Word DOC and DOCX formats with Aspose.PDF for .NET, saving through the Document class with SaveFormat.Doc or with DocSaveOptions set to DocFormat.DocX, on .NET Framework or .NET Core 5 and above, from https://tutorials.aspose.com/pdf/net/conversion-export/convert-pdf-word-aspose-net/ read on 2026-10-16. The tutorial converts PDF files into DOC and DOCX formats and teaches converting PDFs to DOC, advanced conversion options and DOCX conversion with specific configurations. Its prerequisites require an environment supporting .NET Framework or .NET Core 5 and above. It loads the PDF with the Document class and saves with SaveFormat.Doc, or configures DocSaveOptions with Mode Flow, RelativeHorizontalProximity and RecognizeBullets, or sets DocSaveOptions.Format to DocFormat.DocX, then saves. Its FAQ says password protected PDFs can be converted given permissions and passwords. The scope covers only PDF to DOC and DOCX in that direction, and claims no arbitrary formats, no round trip fidelity or losslessness despite the page wording about maintaining layout and structure, no batch behaviour despite performance advice mentioning batching, no PDF/A or PDF/UA conformance and no cross language support, since the source names .NET and its runtimes. Other fields are unchanged.

  135. 7a053ac

    The Aspose.PDF record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as DRM, as unbreakable, as a modern default algorithm claim, as a compliance guarantee, as batch behaviour or as cross language support.
    What it says now
    The job keeps protect and gains the scope encrypting a PDF with Aspose.PDF for .NET using PdfFileSecurity to bind a document and call EncryptFile with separate user and owner passwords, DocumentPrivilege permissions such as printing or modifying content, and AES 256-bit encryption set through KeySize.x256, on .NET Framework or .NET Core, from https://tutorials.aspose.com/pdf/net/security-permissions/encrypt-pdfs-aspose-pdf-net-guide/ read on 2026-10-16. The tutorial encrypts PDFs with Aspose.PDF for .NET, setting user and owner passwords and permissions such as printing only, with AES 256-bit encryption, requiring .NET Framework or .NET Core. It uses PdfFileSecurity from Aspose.Pdf.Facades: BindPdf, then EncryptFile with the user password, owner password, a DocumentPrivilege value, KeySize.x256 and Algorithm.AES, then Save. Its parameter notes define the user password as required for opening, the owner password for changing permissions or removing encryption, DocumentPrivilege.Print as restricting access to printing only, and KeySize.x256 with Algorithm.AES as AES 256-bit. Its FAQ says DocumentPrivilege specifies access levels such as printing or modifying content, and that currently Aspose.PDF supports AES 256-bit encryption for PDFs in .NET applications. The scope claims no DRM, no unbreakability, no algorithm beyond the AES 256-bit named, no legal or compliance guarantee, no batch behaviour, and no cross language support; certificate encryption is not documented on this page and is not claimed, and digital signing and redaction stay separate. Other fields are unchanged.

  136. 7a053ac

    The Aspose.PDF record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a legal validity claim, as trust validation, as timestamping, as recipient certificate encryption, as PAdES, or as covering GUI and batch behaviour.
    What it says now
    The job keeps sign and gains the scope signing a PDF with a digital certificate using PdfFileSignature and a PKCS1 signature with a PKCS#12 certificate file, adding signature fields with FormEditor, setting a custom signature appearance, and verifying signatures with ContainsSignature and VerifySignature, in Aspose.PDF for .NET on a .NET development environment, from https://tutorials.aspose.com/pdf/net/security-permissions/master-pdf-signing-verification-net-aspose-pdf/ read on 2026-10-16. The tutorial covers signing with digital certificates, adding signature fields, verifying signatures and application guidance, and requires Aspose.PDF for .NET in a .NET development environment. Signing loads the document into PdfFileSignature, sets a rectangle and a signature appearance image, creates a PKCS1 signature from a certificate file with a password, and calls Sign with page, reason, name, location, visibility, rectangle and signature. Signature fields are added with FormEditor AddField using FieldType.Signature at page coordinates. Verification uses ContainsSignature and VerifySignature by name or index. The FAQ describes the certificate password as the one securing the PKCS#12 file and confirms custom appearances with images or text. The scope claims no legal validity, no trust validation, no timestamping, no recipient certificate encryption, no PAdES, no algorithm list beyond the PKCS1 signature type and no GUI or batch behaviour. Encryption is the separate protect scope, and redaction is not on this page. Other fields are unchanged.

  137. 7a053ac

    The Aspose.PDF record, in the accessibility job

    What it said
    The accessibility job carried no scope. Earlier Aspose pages mentioned PDF/UA only as a conversion target, which would not have justified the job, so the risk was treating a conversion label as accessibility support.
    What it says now
    The job keeps accessibility and gains the scope creating tagged PDFs and validating them against PDF/UA-1 with Aspose.PDF for .NET, using the tagged content interface to set the title and language and to add paragraph and note elements to the logical structure, and Document.Validate with PdfFormat.PDF_UA_1 to write a validation log, on .NET Core or .NET Framework, from https://tutorials.aspose.com/pdf/net/pdfa-compliance/creating-validating-tagged-pdfs-aspose-net/ read on 2026-10-16. The page differs from the earlier conversion pages: it performs tagging and validation directly. It says accessible PDF creation is essential for screen reader compatibility and meeting standards like PDF/UA, and that the tutorial uses Aspose.PDF for .NET to create tagged PDFs and validate their compliance. It requires .NET Core or .NET Framework. The API creates a Document, takes ITaggedContent from document TaggedContent, sets title and language, creates and appends a ParagraphElement, and creates NoteElements with text and ids appended to the paragraph, through the Aspose.Pdf.LogicalStructure namespace. Validation calls document.Validate with a log file and PdfFormat.PDF_UA_1. The FAQ says a non compliant document should be adjusted after reviewing the validation log, and that other languages are supported through SetLanguage. The scope claims no certification, no automatic remediation, no WCAG breadth beyond PDF/UA, no validation of arbitrary documents, no batch behaviour and no cross language support, and the language attribute is not treated as a language coverage claim. The six settled scopes and every other field are unchanged.

  138. 7a053ac

    The AvePDF record, in the primary compress job, retrieval result

    What it said
    The compress job has no scope and the vendor tool page could not be read. This entry records the render route and the exact failure so the blocker is dated rather than silently repeated.
    What it says now
    Still unresolved, and the tool page is inaccessible to the retrievers available here in a way that is worth recording precisely. The AvePDF compress tool lives at https://avepdf.com/compress-pdf, with a separate Hyper-Compress tool at https://avepdf.com/hyper-compress-pdf. The fetch tool returns HTTP 200 with an empty body. Direct requests with a desktop browser user agent return HTTP 200 with a raw body of exactly zero characters for both pages, and the response contains no __NEXT_DATA__ or __NUXT__ embedded state, no JSON-LD block and no server rendered visible text, so there is nothing to quote and no embedded payload to inspect. The site is therefore fully client side in a way that yields an empty document to any non browser client. A press release and a blog from the vendor parent ORPALIS confirm AvePDF is their free online PDF tool site, but that is marketing about the site rather than documentation of compression behaviour, and ORPALIS product documentation covers different products, so neither is acceptable evidence for this job. The job is therefore left unresolved rather than scoped from the tool name, the homepage or another tool, and it is not removed because nothing contradicts it. The record privacy attributes and the settled convert, protect and redact scopes are untouched. A browser based retrieval or a first party bundle would be needed to settle it.

  139. 7a053ac

    The AvePDF record, in the primary compress job, finding corrected

    What it said
    The previous pass recorded the AvePDF compression page as inaccessible after a plain fetch and direct requests each returned a zero length body, and left the job unresolved on that basis. That conclusion was wrong: the repository already ships a headless browser renderer for exactly this class of client rendered page.
    What it says now
    Corrected and scoped. https://avepdf.com/compress-pdf was rendered with the repository own scripts/render-page.mjs using headless Chrome and a 12000 ms wait, giving 3,614 characters of visible text from a 417,974 byte DOM in about 14 seconds. The rendered page says Compress and reduce the size of your PDF, states a Max file size of 128 MB, says to choose the file compression level desired low, medium, high or very high and click Compress, then download or save the result. It is Powered by Nutrient PDF Optimization utilizing the GdPicture.NET SDK, and a banner says 2 free actions remain before AvePDF Premium is needed. The page contains no occurrence of lossless and no guarantee wording. The job therefore gains the scope compressing a PDF in the browser on AvePDF by uploading a file of up to 128 MB and choosing a low, medium, high or very high compression level before downloading or saving the compressed PDF, powered by Nutrient PDF Optimization and the GdPicture MRC compression SDK, with free use limited to a small number of actions before Premium is required. Its privacy notice says uploads stay accessible for up to 30 minutes and that free tier uploads are retained and used to improve services, which is consistent with the record privacy attributes, so those were left unchanged. Lesson recorded: a zero length response is not evidence that a vendor does not document something, and the renderer must be tried before any such finding is recorded.

  140. 7a053ac

    The AvePDF record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as preserving bookmarks or forms, as lossless, as general batch processing, as supporting arbitrary formats, or as requiring a paid plan beyond the stated free action limit.
    What it says now
    The job keeps merge and gains the scope merging multiple PDFs into one file in the browser on AvePDF by uploading a first PDF and then more files, each up to 128 MB, checking the file order and reorganising it before clicking Merge and downloading or saving the merged PDF, powered by the Nutrient PDF Editor and GdPicture.NET SDK, with free use limited to a small number of actions before Premium is required, from https://avepdf.com/merge-pdf rendered on 2026-10-16 with the repository own scripts/render-page.mjs, headless Chrome 155, a 12000 ms wait, roughly 14 seconds and a 3,179 character visible text from a 402,016 byte DOM. The rendered page says Merge multiple PDFs into one file, notes a Max file size of 128 MB, says the tool is powered by Nutrient PDF Editor utilizing GdPicture.NET SDK, and its how-to section instructs dropping or uploading a first PDF, uploading more files once the first is processed, checking the file order and reorganising by moving files left or right, clicking Merge, then downloading or saving the merged PDF. A banner says 2 free actions remain before Premium. The page contains no occurrence of bookmark or lossless. The scope therefore claims no bookmark or form preservation, no losslessness, no general batch support, no arbitrary formats and no paid plan beyond the free action limit. The privacy notice matches the compression page and the record privacy attributes were unchanged. The settled compress, convert and protect scopes and every other field are unchanged.

  141. 7a053ac

    The AvePDF record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as an accuracy claim, as handwriting support, as editable output, or as language breadth beyond the six languages the tool lists.
    What it says now
    The job keeps ocr and gains the scope running OCR in the browser on AvePDF on files in more than 100 formats up to 128 MB to generate a searchable PDF, choosing OCR languages from English, French, German, Spanish, Italian or Portuguese, powered by Nutrient OCR and the GdPicture.NET SDK, with free use limited to a small number of actions before Premium is required, from https://avepdf.com/pdf-ocr rendered on 2026-10-16 with the repository own scripts/render-page.mjs, headless Chrome 155, a 12000 ms wait and a 4,093 character visible text. The page says OCR PDF, accurately recognize text and make your PDF searchable, notes a Max file size of 128 MB, and is powered by Nutrient OCR utilising GdPicture.NET SDK. Its how-to section names more than 100 supported formats, the six OCR languages English, French, German, Spanish, Italian and Portuguese, a Start OCR button, and downloading the result as PDF OCR. A banner says 2 free actions remain. Background text explains OCR and warns that OCR is not magical and some documents are tough to decipher. The page contains no occurrence of accuracy, editable or handwriting. The scope therefore claims no accuracy, no handwriting, no editable output and no language breadth beyond the six named, and carries searchable output only because the page states it as the tool purpose. Retrieval note: the /ocr-pdf slug returned a 404 that still served the full site chrome, so the real /pdf-ocr route was taken from the rendered DOM anchors. Privacy attributes unchanged. The settled compress, merge, convert and protect scopes and every other field are unchanged.

  142. 7a053ac

    The AvePDF record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as legal validity or eIDAS compliance, as PAdES, as trust validation, as an audit trail, as a certificate-based digital signature profile, or as general batch signing.
    What it says now
    The job keeps sign and gains the scope signing a PDF in the browser on AvePDF by drawing a signature with the cursor or importing one from the device, placing it by drawing a box on the document, optionally importing a certificate with its password for an additional level of security, and downloading or saving the signed PDF, powered by Nutrient PDF Signature and the GdPicture.NET SDK, with free use limited to a small number of actions before Premium is required, from https://avepdf.com/esign-pdf rendered on 2026-10-16 with the repository own scripts/render-page.mjs, headless Chrome 155 and a 12000 ms wait, producing 5,302 characters of visible text. The route was recovered from DOM anchors on a rendered homepage, where eSign PDF resolves to /esign-pdf, rather than by guessing. The page says Sign your PDF online, notes a Max file size of 128 MB, and is powered by Nutrient PDF Signature utilising GdPicture.NET SDK. Its how-to section describes the Signature button, drawing or importing a signature, an optional Import certificate step with a password for additional security, saving, then drawing a box to place the signature and saving again, followed by download or cloud save. A banner says 2 free actions remain. A background section summarises eIDAS signature types and national laws. The page contains no occurrence of PAdES or audit. The scope claims no legal validity or eIDAS compliance, no PAdES, no trust validation, no audit trail, no certificate-based digital signature profile and no batch behaviour, and limits paid wording to the stated free action limit. The settled compress, merge, ocr, convert and protect scopes, the privacy attributes and every other field are unchanged.

  143. 7a053ac

    The AvePDF record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as editing existing document text, as metadata editing, as reflow, as OCR or redaction, or as lossless.
    What it says now
    The job keeps edit and gains the scope editing a PDF in the browser on AvePDF with the Online PDF Editor by adding text, images and other items drawn directly on the page and then selecting and moving or recolouring them, and rotating pages, on files up to 128 MB, powered by Nutrient PDF Editor and the GdPicture.NET SDK, with free use limited to a small number of actions before Premium is required, from https://avepdf.com/pdf-edit rendered on 2026-10-16 with the repository own scripts/render-page.mjs, headless Chrome 155 and a 12000 ms wait, renderer exit 0. The route was recovered from DOM anchors on a rendered homepage. The editor page says Easily edit your PDF online, rotate pages, add text and images, and more, notes a Max file size of 128 MB, and is powered by Nutrient PDF Editor utilising GdPicture.NET SDK. Its how-to adds content by clicking an item button and drawing on the document, selects an added item with the Selection tool to move it or edit its colours, then downloads the edited PDF. A second tool at /edit-pdf-meta edits only PDF metadata such as file name, title, author, subject, keywords and dates, and was deliberately not folded in because metadata editing is not page content editing. The page contains no reflow, OCR or redaction wording for this tool. The scope claims no editing of existing document text or content, because the page describes adding items and editing the added item; no reflow, OCR, redaction, losslessness or batch behaviour; and no metadata editing. Paid wording is limited to the stated free action limit. Retrieval note: a 404 page on this site serves the full chrome, and the anchors of any rendered page are the reliable route directory. The settled compress, merge, ocr, sign, convert and protect scopes, the privacy attributes and every other field are unchanged.

  144. 7a053ac

    The Azure AI Document Intelligence record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as an accuracy claim, as covering every model and the whole prebuilt catalogue, as language breadth, as including searchable PDF output or OCR add-ons, or as covering container deployment.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting text, tables, document structure and key-value fields from documents with the cloud-based Azure AI Document Intelligence service, using the Read, Layout, prebuilt or custom models, with fields returned as strongly typed values, through the REST API, C#, Python, Java and JavaScript SDKs or Document Intelligence Studio, from https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/overview read on 2026-10-16, the official Microsoft Learn overview for the current v4.0 release. The page describes a cloud-based Foundry Tools service for intelligent document processing; document analysis models that extract text from forms and documents and return structured content, with Read for printed and handwritten text and Layout for text, tables and structure; prebuilt models for named document classes; custom neural, template, composed and classifier models; and strongly typed field values covering types such as string, number, integer, date, time, phoneNumber, currency and address. It lists Document Intelligence Studio, the REST API and the C#, Python, Java and JavaScript SDKs as development options, and separately lists add-on capabilities including searchable PDF, formula, font, barcode, query fields and key value pairs. The scope claims no accuracy, no complete model or prebuilt coverage, no language breadth, no file limits and no container deployment, and it deliberately excludes the searchable PDF and OCR add-ons so the separate OCR job and the conversion job stay distinct. Other fields are unchanged.

  145. 7a053ac

    The Azure AI Document Intelligence record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as an accuracy claim, as handwriting quality, as language breadth or coverage, as arbitrary input types, or as including searchable PDF output.
    What it says now
    The job keeps ocr and gains the scope extracting printed and handwritten text from PDF, JPEG, PNG, BMP, TIFF and HEIF inputs with the Read and Layout models of the cloud Azure AI Document Intelligence service, with optional add-ons for high resolution extraction, font properties, formulas, barcodes and language detection, from https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/concept/add-on-capabilities read on 2026-10-16, the official Microsoft Learn add-on capability page for v4.0, last updated 2026-08-15. The page lists the add-ons as ocrHighResolution, formulas, styleFont, barcodes, languages, Searchable PDF support, queryFields and keyValuePairs, marks barcode extraction, language detection, key value pairs and searchable PDF as free and font, formula, high resolution and query fields as add-ons, and states the supported inputs as PDF plus JPEG, JPG, PNG, BMP, TIFF and HEIF, with Microsoft Office files not supported. It describes high resolution extraction for small text in large engineering documents and A1 to A3 sizes, formula extraction with LaTeX values and a hard coded confidence, font property extraction with font family, style, weight, colour and background colour, barcode extraction with a hard coded confidence and a named barcode type list, and language detection predicting a primary language per line with confidence. The scope claims no accuracy, no handwriting quality, no language breadth since only detection is documented, no inputs beyond the explicit list, no batch behaviour, no containers and no product wide limits. Searchable PDF is documented but kept out of the OCR scope because it is an output tied to the Read model and belongs with the untouched convert job. The primary extract scope and other fields are unchanged.

  146. 7a053ac

    The Azure AI Document Intelligence record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as arbitrary format conversion, as Office or image to PDF conversion, as round trip fidelity, as PDF/A output, as batch behaviour or as an accuracy claim.
    What it says now
    The job keeps convert and gains the scope converting an analog or image-only PDF into a PDF with an embedded searchable text layer through the Searchable PDF capability of the cloud Azure AI Document Intelligence service, using the prebuilt-read model with the pdf output option, from the Searchable PDF section of https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/concept/add-on-capabilities read on 2026-10-16. That section says the capability enables you to convert an analog PDF, such as scanned-image PDF files, to a PDF with embedded text, that the embedded text enables deep text search by overlaying detected text entities on the image files, and that currently only prebuilt-read supports it, with the modelId set accordingly and included in the 2024-11-30 GA model at no usage cost for general PDF consumption. The usage section posts to prebuilt-read analyze with output=pdf and retrieves the result as application/pdf. The vendor wording is a genuine conversion statement rather than a label, so the job is scoped, but because the operation is PDF in and PDF out the scope says so explicitly so it cannot be mistaken for format conversion. The scope claims no arbitrary format conversion, no Office or image to PDF conversion which the same page says is unsupported, no round trip fidelity, no PDF/A, no batch behaviour, no accuracy and no cross language support. The settled extract and ocr scopes and every other field are unchanged.

  147. 7a053ac

    The Bluebeam Revu record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as markup editing, as reflow, as OCR, as form editing, as redaction or as collaboration.
    What it says now
    The job keeps edit, remains primary, and gains the scope adding, replacing or deleting existing text in a PDF with the Edit Text tool in Bluebeam Revu, which matches the font, style and size of the text around it but will not wrap added text, and cannot be used on digitally signed or certified PDFs, from https://support.bluebeam.com/nl/online-help/revu20/Content/RevuHelp/Menus/Tools/Markup/PDF-Content/Text-Editing--MTV.htm read on 2026-10-16, the official Revu 20 online help page Editing and Reviewing Text. The page separates editing from reviewing: Editing Text says Edit Text adds, replaces or deletes existing text in a PDF, while Review Text is described as marking text for edits without changing the text in the PDF, with both review sections repeating that they do not change the actual text. Other text markups cover underline, squiggly and strikethrough. The scope is taken from the editing section only, and the markup and review tools are deliberately not included. The vendor own boundaries are carried in the scope: added text will not wrap and the tool is recommended for minor changes, and Edit Text cannot be used on a digitally signed or certified PDF. Revu recognises and matches the surrounding font, style and size, prompting a font change when the document uses a font not on the computer. The scope claims no reflow, OCR, form editing, redaction permanence, collaboration or format conversion. The version boundary is the Revu 20 help set. The settled redaction entitlement entry, pricing, privacy, platforms and every other field are unchanged, and this evidence exposed no contradiction with them.

  148. 7a053ac

    The Bluebeam Revu record, in the view job

    What it said
    The view job carried no scope, so it could be read as including editing, printing, form filling, collaboration, accessibility, free licensing or cross platform parity.
    What it says now
    The job keeps view and gains the scope viewing PDFs in Bluebeam Revu, with multiple documents open in document tabs and a Document Access menu, view rotation that changes only the display and is not saved to the file, and a full-screen Presentation mode, with Revu positioned for viewing multiple large drawings, from https://support.bluebeam.com/user-manual/view-pdfs.html read on 2026-10-16, the View PDFs page of the official Revu User Manual. The page says Revu is tuned for viewing multiple large drawings easily and efficiently, that multiple PDFs can be open in separate tabs and as many as resources allow, that a Document Access menu lists open PDFs with a checkbox for the active one, that Rotate View changes only the display and will not be saved with the PDF while Rotate Pages is for permanent rotation, and that Presentation mode shows the current PDF in a slide-show style full-screen view with keyboard navigation. The scope carries the display only nature of Rotate View because the vendor contrasts it with Rotate Pages, which keeps it from reading as editing. It claims no free licensing, no cross platform parity, no annotation or editing, no printing, no form filling, no offline behaviour, no accessibility and no collaboration. The surface is the Revu desktop user manual and no platform claim is drawn from it. The settled edit scope, the redaction entitlement entry, pricing, privacy, platforms and every other field are unchanged, and no contradiction with them was exposed. Retrieval note: a locale-prefixed manual URL returned 404 and the canonical locale-free user-manual path taken from the support footer worked.

  149. 7a053ac

    The Bluebeam Revu record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a signature appearance rather than cryptographic signing, or as implying legal validity, eIDAS or PAdES compliance, an audit trail, remote signing, certificate authority trust validation or a general e-signature workflow.
    What it says now
    The job keeps sign and gains the scope digitally signing and certifying PDFs in Bluebeam Revu, with digital signature fields, digital IDs from self-signed certificates or certificates purchased through third-party certificate authorities, signature validation and the Signatures panel, validating and signing against the Windows Certificate Store and PKCS #12 standards, with Adobe CDS signatures also supported, from https://support.bluebeam.com/user-manual/menus/tools/digital-signatures.html read on 2026-10-16, the Digital signatures page of the official Revu User Manual. The page says Revu supports adding digital signature fields, creating and validating signatures from self-signed certificates for certification within organisations or among trusted parties, and using commercially available certificates purchased through trusted third-party Certificate Authorities; that Revu validates and signs documents based on the Windows Certificate Store and PKCS #12 standards and also supports Adobe CDS signatures; and it links procedures for signature fields, self-signed digital IDs, importing a digital ID, certifying a document, digitally signing a document, validating a signed document, importing and exporting a trusted identity certificate, and customising signature appearance. It also describes the Signatures panel holding certification and signature information for any certified or digitally signed PDF. The scope claims no legal validity, no eIDAS or PAdES, no audit trail, no remote signing and no general e-signature workflow, and it does not assert certificate authority trust validation, only that certificates can be purchased through third-party authorities as the page states. It is built on the signing page rather than any markup or signature image feature. The settled edit, view and protect scopes, the redaction entitlement entry, pricing, privacy, platforms and every other field are unchanged, and no contradiction was exposed. Retrieval note: a static Revu 20 help path for signatures returned 404, and the canonical page was found through the manual dashboard link map at the locale-free user-manual path.

  150. 7a053ac

    The borb record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as arbitrary text editing or find and replace, as annotation mutation, as reflow, as OCR or redaction, or as a commercial feature tier.
    What it says now
    The job keeps edit and gains the scope editing a PDF with the borb Python library, editing document metadata and appending pages to the page tree, with borb described as a Python library for creating and manipulating PDFs and dual licensed under AGPL and a commercial licence, from https://raw.githubusercontent.com/jorisschellekens/borb/master/README.md read on 2026-10-16. The README describes borb as a Python library for creating and manipulating PDFs and a pure Python solution for PDF document management that reads, writes and manipulates PDFs, and its feature list names PDF Metadata Management described as reading and editing, text and image extraction, adding annotations, content manipulation by adding text images tables and lists, and page layout management. Its Hello World example creates a Document and Page, calls append_page, builds a page layout, appends a Paragraph and writes the PDF. The README states dual licensing under AGPL and a commercial licence with commercial use requiring a paid licence, and its badges list Python 3.10, 3.11 and 3.12. The scope therefore claims no arbitrary text editing or find and replace, no reflow, no annotation mutation since the source says adding annotations, no OCR, no redaction permanence, no compression or encryption, and no add-on capability is claimed; the README names only the dual licence, while the record's licensing field documents paid closed-source add-ons including borb_redact, borb_ocr, borb_markdown_to_pdf and borb_find_replace. The settled convert and merge scopes, the licensing, telemetry and privacy findings and the redaction uncertainty are unchanged and no contradiction was exposed.

  151. 7a053ac

    The Camelot record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as generic text extraction, as supporting image-only PDFs on a plain install, as an accuracy guarantee, or as a benchmark claim.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting tables from text-based PDFs with the Camelot Python library, returning each table as a pandas DataFrame and exporting it to CSV, JSON, Excel, HTML, Markdown or SQLite, taking a file path, URL, raw bytes or binary file-like object as input, from https://raw.githubusercontent.com/camelot-dev/camelot/master/README.md read on 2026-10-16. The README describes Camelot as a Python library that helps extract tables from PDFs, shows camelot.read_pdf returning a TableList with tables[0].df as a pandas DataFrame, and documents export and per-table methods for csv, json, excel, html, markdown and sqlite, and flexible input from a file path, URL, raw bytes or binary file-like object. It also states that the built-in parsers need a text-based PDF and that scanned or image-only PDFs need the optional ml and ocr extras. The scope claims no generic text extraction, no editing or conversion back to PDF, no accuracy guarantee, no benchmark results and no image-only support for the base install. Parsers, the ml and OCR extras and the quality metrics are left out as separate concerns, and no parser default is asserted because the README calls lattice the default while flavor auto picks lattice or network. The MIT licence, dependencies and other settled attributes and the other jobs are unchanged.

  152. 7a053ac

    The CamScanner record, in the primary ocr job

    What it said
    The ocr job carried no scope, so it could be read as producing a searchable or editable PDF, as an accuracy or handwriting claim, as language coverage, as a free feature, or as running locally because scanning happens on the device.
    What it says now
    The job keeps ocr, remains primary, and gains the scope recognizing text in images or PDFs with the OCR feature of the CamScanner scanner app, extracting that text for later searching, editing or sharing, and searching for images by their recognized content, from the official Google Play listing for com.intsig.camscanner at https://play.google.com/store/apps/details?id=com.intsig.camscanner&hl=en_US, read on 2026-10-16 through the repository own headless renderer because a plain fetch returned an empty body. The listing Extract Text entry says the optical character recognition feature enables you to recognize text in images or PDFs and that you can extract the text for later searching, editing, or sharing. The description says the app recognizes text automatically (OCR), and the Quick Search entry says the OCR feature enables you to search for images based on their content. The scope claims no accuracy, no handwriting, no language coverage and no searchable or editable PDF output, since the listing speaks of extracting the text rather than producing a text layered PDF. It claims no batch behaviour and no arbitrary conversion. No free versus paid OCR boundary is stated, the subscription text offering only unlimited access to all features, so no per feature gate is claimed. Nothing states whether recognition is on the device or on a server, so no local processing claim is made despite scanning being presented as happening on the mobile device. The privacy and pricing fields, the settled convert scope, the other jobs and every other field are unchanged, and no contradiction was exposed.

  153. 7a053ac

    The CamScanner record, in the compress job, retrieval result

    What it said
    The compress job has no scope. This entry records what the official sources do and do not say about compression, so the blocker is dated rather than silently repeated.
    What it says now
    Still unresolved. Two first party sources were read on 2026-10-16 through the repository own headless renderer, each with a 12000 ms wait and renderer exit 0. The official Google Play listing for com.intsig.camscanner produced 8,407 characters of visible text and contains no occurrence of compress, compression, file size, smaller, downsizing or lossless, and no sentence describing compression at all; its only related wording is Optimize Scan Quality, which speaks of smart cropping and auto enhancing rather than compression. The official site at https://www.camscanner.com/ produced 8,596 characters of visible text and mentions compress exactly once, inside a customer testimonial rather than product documentation: a teacher is quoted saying adding annotations, and merging, splitting, and compressing files are all a breeze. The page has no compression controls, no input or output statement, no size or quality settings, no service surface, no file size figures and no lossless claim. A testimonial naming a feature is not vendor documentation of what the feature accepts, produces or limits, so the job is left unresolved rather than scoped from it. It is not removed because nothing contradicts it. A dedicated compression or toolbox help page, or a first party manual entry, would be needed to settle it.

  154. 7a053ac

    The CamScanner record, in the edit job, retrieval result and exclusion

    What it said
    The edit job has no scope. This entry records that the official sources describe annotation markup rather than content editing, so the job is excluded on evidence rather than left ambiguous.
    What it says now
    Still unresolved, and the exclusion is specific. Three first party sources were read on 2026-10-16 through the repository own headless renderer, each 12000 ms wait with renderer exit 0. The official Google Play listing for com.intsig.camscanner produced 8,407 characters of visible text, and its editing feature is described as follows: Advanced Document Editing, make annotations on documents using a full set of editing tools in this PDF scanner. The operation it names is making annotations, which is markup laid over a document rather than changing existing text, images or document structure. Its other edit mentions are not PDF content editing either: one says you can extract the text for later searching, editing, or sharing, which edits the extracted text, and another says you can view, edit and share any document with the app on the go, which names no operation at all. A separate app in the listing, CS PDF Reader with PDF Editor, is a different product and is not evidence for this record. The web toolbox was checked too: https://www.camscanner.com/edit-pdf returns a 404 page of 238 characters, and although the homepage navigation carries a link labelled Edit PDF Online it points to https://www.camscanner.com/merge-pdf, whose 7,831 characters of visible text describe merging and choosing file order, with its only rearrange wording inside a customer testimonial. So neither surface documents editing existing PDF content. The job therefore stays unresolved rather than being scoped from the generic phrase PDF editing or from the editing tools label; it is not removed because nothing contradicts it. A dedicated editing or help page describing an operation on existing content would be needed to settle it.

  155. 7a053ac

    The CamScanner record, in the sign job, retrieval result and exclusion

    What it said
    The sign job has no scope. This entry records that the official signing page documents a drawn or uploaded signature image rather than a cryptographic digital signature, so the job is excluded on evidence rather than left ambiguous.
    What it says now
    Still unresolved, and the exclusion is specific. The official CamScanner web toolbox page at https://www.camscanner.com/sign-pdf was read on 2026-10-16 through the repository own headless renderer, 12000 ms wait, renderer exit 0, giving 1,991 characters of visible text. The page is titled Sign PDF Online, Fill and Sign PDF for Free, and says add your signature to PDF documents online, then save the signed file for sharing or recordkeeping. Its Your Signature Your Way section says draw your signature by hand or upload an image of it, and add seals, logos, or cross-page seals when your document needs more than a signature. It also documents signing for other people, saying sign a document on your own or send it to other people for their signatures and set a due date, and states that the service is cloud based, saying there are many servers in the cloud which do nothing else but add your signatures to PDF files. The page contains no occurrence of certificate, digital signature, digital certificate, PAdES, eIDAS, identity verification or audit trail. The operation documented is therefore placing a signature image or seal on a document, which is markup rather than a cryptographic signature, and the multi-party invite and due date wording describes a signing request workflow without stating any cryptographic mechanism. A plain English reading of sign could include that workflow, but the catalog job is being held to a cryptographic standard consistent with the Bluebeam entry, so the job stays unresolved rather than being scoped from signing language alone. It is not removed because nothing contradicts it. A first party page documenting certificates or digital signatures, or a decision that a drawn signature satisfies the job, would settle it.

  156. 7a053ac

    The CamScanner record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as including an encryption algorithm, permissions restrictions, key management, local processing, or a free versus paid gate, or conversely as a watermark.
    What it says now
    The job keeps protect and gains the scope adding an opening password to a PDF with the CamScanner web toolbox Protect PDF tool, so recipients must enter the password before viewing the document, producing a protected PDF file to share with the password sent separately, with processing on CamScanner cloud servers, from https://www.camscanner.com/protect-pdf rendered on 2026-10-16 with the repository own headless renderer, 12000 ms wait, renderer exit 0, giving 2,143 characters of visible text. The page says to add an opening password so recipients need to enter it before viewing the PDF, and states that the password controls access to the document rather than simply placing a visible label on its pages, which distinguishes it from a watermark. It advises saving the protected PDF for sharing and sending the password through a separate channel, and says files are processed by cloud servers that add password protection to PDFs. The page contains no mention of an encryption algorithm, permissions, printing or copying restrictions, key management or certification; its only encryption wording is 256-bit SSL, which describes the service transport rather than the PDF, so no PDF encryption algorithm is claimed. The page title says the tool is free, but no plan gate is stated, so no free versus paid boundary is claimed, and no local processing is claimed because the vendor says processing is cloud based. The Play listing separately says a password can be set for viewing and for a document download link, which is access control around the app and link rather than PDF protection, and was not folded into the scope. The settled ocr and convert scopes, the compression, edit and sign blockers and every other field are unchanged.

  157. 7a053ac

    The Chunkr record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as including the Parse feature HTML and Markdown output, as layout analysis or OCR, as an accuracy or benchmark claim, as spreadsheet credit rates or pricing, or as making a managed versus self hosted claim.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting structured data from documents with the Chunkr Extract feature, filling a JSON Schema you define and returning the result in that schema shape with a source citation and a confidence score for each extracted value, accepting a raw document from a URL, file upload or base64, or an existing parse task, from https://docs.chunkr.ai/pages/features/extract/overview.md read on 2026-10-16, reached through the official index at https://docs.chunkr.ai/llms.txt. The page says Extract transforms parsed documents into structured data based on your defined schema and that each extracted value comes with granular citations and confidence; its key features are schema driven extraction using JSON Schema, granular citations giving precise source references to the original document location, per field confidence scoring, flexible input from existing parse tasks raw documents or remote URLs, and intelligent field mapping. The how it works section gives the input as a raw document URL, file upload or base64, or a reference to an existing parse task, and the output as the data in the exact schema format with enriched metadata; schemas are defined with Pydantic or Zod and passed as JSON Schema to an extract task. The scope claims no accuracy, benchmark or human level accuracy, no handwriting, layout analysis or OCR, which belong to Parse, and no HTML or Markdown output, which is also Parse rather than Extract JSON. It claims no spreadsheet credit rates, retention, compliance, pricing or free allowance, and makes no managed versus self hosted claim, since that boundary is a record level question whose existing contradictions are preserved. Note for a later pass, not folded into this scope: the docs welcome page says Chunkr is joining Extend and that the hosted Chunkr API shuts down on 2026-11-15, which is a hosting question for the record rather than a job scope question. The Parse job and every other field are unchanged.

  158. 7a053ac

    The Chunkr record, in watchOutFor and internalNotes, the hosted API lifecycle

    What it said
    The record described Chunkr as a product you can use as a managed API or self host, with published monthly pricing and no end date anywhere. Chunkr documentation now announces that the hosted Chunkr API shuts down on 2026-11-15, so a reader could plan a hosted integration or budget the monthly plans without knowing the managed edition is closing.
    What it says now
    Current. watchOutFor now says the hosted API has a published end date: the documentation states Chunkr is being acquired by Extend, the hosted Chunkr API stays fully live through 14 November 2026 and shuts down on 15 November 2026, and a migration guide covers moving a Parse or Extract integration. It says the shutdown applies to the hosted API only, because the same documentation says the open-source version stays up, remains available under its existing licence and needs nothing done if run with Docker Compose, so a self-hosted deployment is unaffected. It adds that pricing the monthly plans should follow a read of the migration guide, since an API key does not carry over and documents and task outputs are not migrated. The three sources, all first party and all read 2026-10-16, are https://docs.chunkr.ai/pages/get-started/welcome (joining Extend, hosted API shuts down on November 15, 2026), https://docs.chunkr.ai/pages/migrate-to-extend/overview (being acquired by Extend, stays fully live through November 14, 2026, shuts down on November 15, 2026) and https://docs.chunkr.ai/pages/migrate-to-extend/faq, which settles the scope by saying the open-source version stays up, remains available under its existing license, needs nothing done under Docker Compose, and is separate from the hosted API so the migration guide does not apply to it. status stays published and discontinued stays false because the date is in the future and only one of two editions ends; the self-hosted edition, AGPL licence, platforms, editions, extract scope, pricing, claims, summary and tagline are unchanged. internalNotes carries the same evidence with the wording differences between the pages recorded rather than resolved.

  159. 7a053ac

    The CommonLook PDF record, in the primary accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as a guarantee of legal compliance, as automatic correctness, as OCR or generic PDF editing, as batch processing, or as an AI accuracy claim.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope remediating an uploaded PDF for accessibility with CommonLook PDF, which uses AI driven tools to tag documents automatically and offers simplified or advanced editing for the remediation work, then validating the result against the PDF/UA, WCAG and HHS standards and producing a downloadable validation report listing passed, failed and warning issues, available through a web browser or as a desktop download, from https://allyant.com/pdf-accessibility-software-solutions/commonlook-pdf/ read on 2026-10-16, the official Allyant product page for CommonLook PDF. The page gives four steps, upload a PDF, select simple or advanced editing, follow prompts for remediation, download accessible files; it says AI driven tools automatically tag documents and that users choose an advanced or simplified editor based on technical understanding and PDF complexity; and a Conformance and Compliance Validation section names PDF/UA, WCAG and HHS, says each accessible PDF comes with a unique downloadable Validation Report capturing passed, failed and warning issue counts, and offers web based access or a desktop download. The scope explicitly does not claim legal compliance or correctness: the page says the software enables organisations to meet ADA Title II, Section 508, HHS and EAA requirements, but that is a marketing claim about customer outcomes, so the scope names only the standards validated against and the report produced. No automatic correctness, OCR, generic PDF editing, batch processing or AI accuracy is claimed. The settled quote only pricing, the mixed web and Acrobat delivery model, privacy and data residency and every other field are unchanged.

  160. 7a053ac

    The CommonLook PDF record, in the edit job, retrieval result and exclusion

    What it said
    The edit job has no scope, so it could be read as general PDF content editing. The official product page describes an editor, which makes that reading more likely than the label alone would.
    What it says now
    Still unresolved, and the exclusion is specific. The official Allyant product page for CommonLook PDF at https://allyant.com/pdf-accessibility-software-solutions/commonlook-pdf/ was read on 2026-10-16 through the repository own headless renderer, 12000 ms wait, renderer exit 0, giving 5,362 characters of visible text. Its Standout Software Features define what the editors do: Advanced or Simplified Editing lets a user choose a more advanced editor for practitioners or a simplified tool based on experience and PDF complexity; AI-Driven Automated Tagging and Remediation accelerates tagging and supports the remediation of tables, lists, forms and other document elements; and Automated Bookmark Generation creates bookmarks based on structural heading levels. The workflow is upload a PDF, select simple or advanced editing, follow prompts for remediation, download accessible files. The same rendered text contains no occurrence at all of text, image, layout, OCR, redaction, annotation, conversion, compression, reflow or font as something the editor changes; every occurrence of editing refers to those remediation editing paths. The operations documented are therefore changes to tag structure, elements, reading navigability and bookmarks, which is accessibility remediation and is already carried by the accessibility job on this record, not general editing of existing page content. Scoping edit from this page would duplicate the accessibility scope under a second job and would tell a reader the tool edits document content, which it does not claim. The job therefore stays unresolved rather than being scoped or removed, since nothing contradicts it; a source describing content editing would settle it.

  161. 7a053ac

    The Datalogics Adobe PDF Library record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a signature appearance rather than cryptographic signing, as including certification, as a claim of legal validity or eIDAS compliance, as trust chain validation, remote signing or an audit trail, or as applying to every language binding.
    What it says now
    The job keeps sign and gains the scope digitally signing a PDF with the Datalogics Adobe PDF Library SignDoc class, setting the signature field, credential data and digest category and choosing among the documented signature types CMS, RFC3161 trusted timestamp and PAdES baseline B-T, as documented in the APDFL 18 .NET Framework API reference, from https://docs.datalogics.com/apdfl18/dotnetframework/apdfl18.0.5plusp2d/class/signdoc/ and https://docs.datalogics.com/apdfl18/dotnetframework/apdfl18.0.5plusp2d/enum/signaturetype/ read on 2026-10-16. The SignDoc page says the class is the implementation of the PDF SignDoc function and that the SignDoc object is used to digitally sign a document; its properties include CredentialDataFormat, DigestCategory, DocSignType of type SignatureType, FieldID of type SignatureFieldID, FieldName and OutputPath, which is what makes this a credential based signing API rather than an appearance feature. The SignatureType enumeration defines CMS as a digital signature based on the Cryptographic Message Syntax standard whose signatures contain embedded timestamps per RFC3161, RFC3161 as a trusted timestamp based on the Time-Stamp Protocol, and PADES as a PAdES baseline B-T digital signature based on the ETSI EN 319 142-1 European Standard. The scope claims no legal validity, no eIDAS compliance, no trust chain validation, no remote signing, no audit trail and no redaction. Certification is not claimed, since it appears only as a neighbouring enum name in the navigation rather than as a documented operation on these pages. The language boundary is stated because the reference read is the APDFL 18 .NET Framework one and the other bindings were not read in this pass. The settled edit, protect, OCR, merge, extract and compress scopes, the intentionally unverified redaction field, pricing, licensing, platforms and every other attribute are unchanged.

  162. 7a053ac

    The Datalogics Adobe PDF Library record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format conversion, as an XPS path, as a guarantee that output conforms to PDF/A or PDF/X, or as including fidelity, OCR, editing, compression or batch behaviour.
    What it says now
    The job keeps convert and gains the scope converting a document to the PDF/A or PDF/X standard with the Datalogics Adobe PDF Library, using PDFAConvertParams and PDFXConvertParams to control the conversion, including options to remove all annotations and actions from the output, rasterize pages when fonts cannot be embedded, abort when XFA form content is present and either ignore or validate conversion errors, as documented in the APDFL 18 .NET Framework API reference, from https://docs.datalogics.com/apdfl18/dotnetframework/apdfl18.0.5plusp2d/class/pdfaconvertparams/ and https://docs.datalogics.com/apdfl18/dotnetframework/apdfl18.0.5plusp2d/class/pdfxconvertparams/ read on 2026-10-16. PDFAConvertParams is described as parameters for converting a document to PDF/A compliance that control how a Document is converted to the PDF/A standard, listing RemoveAllAnnotations, IgnoreFontErrors which rasterizes the page when fonts cannot be embedded, AbortIfXFAIsPresent, NoValidationErrors, RasterizationResolution defaulting to 72 DPI, and ValidateImplementationLimitsOfDocument. PDFXConvertParams is described as parameters for converting a document to PDF/X compliance that control how a Document is converted to the PDF/X standard, with a constructor taking RemoveAllAnnotations and AbortIfXFAIsPresent. The scope claims no universal format coverage: XPSConvertParams and the result classes are visible in the same navigation but their pages were not read, so XPS and a general format matrix are not claimed. It claims no fidelity or conformance guarantee, and the NoValidationErrors option is precisely why, since a caller can ignore validation errors rather than being promised that output conforms. No OCR, editing, compression, batch behaviour or licensing wording is claimed, and the language boundary is stated as the APDFL 18 .NET Framework reference because that is the reference read. The settled edit, protect, sign, OCR, merge, extract and compress scopes, the intentionally unverified redaction field, platforms, pricing and licensing are unchanged.

  163. 7a053ac

    The DocHub record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a drawn signature or signature field rather than a certificate based signature, as legal validity, as eIDAS or PAdES compliance, as identity verification or as a plan gated feature.
    What it says now
    The job keeps sign and gains the scope downloading a PDF from DocHub as a digitally certified copy, or receiving one automatically from a Sign Request, where the PDF carries an embedded X.509 PKI digital signature issued by DocHub that indicates the document has not been modified since signing and proves its authenticity, is LTV enabled, and may embed signer details such as name, email and geolocation in the signature reason, from https://help.dochub.com/knowledge-base/document-signing-and-sign-requests/certified-documents-and-verified-signatures-adobe-acrobat-valid-esignatures read on 2026-10-16. The article says signed documents may be downloaded as a PDF with an embedded digital signature that ensures the document has not been modified and proves its authenticity, that certified copies are produced by checking the certify document option in the download modal, and that PDFs from a Sign Request are automatically digitally signed and certified including all revisions and the audit trail. It says DocHub uses an X.509 PKI certificate to certify the PDF, that signatures are LTV enabled so validity does not expire in Acrobat, and that the signature may embed signer name, email and geolocation, appearing as Signed by DocHub LLC. The scope is taken from that certificate article rather than the drawn or typed signature workflow, and it says the certificate is DocHub issued rather than signer supplied, which is what the source describes; the first party blog on electronic signatures draws the same line, calling a digital signature certificate based and distinct from a visible signature element. No legal validity, eIDAS compliance, PAdES profile, identity verification, certificate authority trust, remote signing security or plan requirement is claimed. The article does assert adherence to the intent of the US Federal ESIGN Act, but that is a vendor legal claim rather than a technical property and is not carried into the scope. The settled attributes, jobs, pricing, privacy and platform findings are unchanged.

  164. 7a053ac

    The DocHub record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as preserving bookmarks, as lossless, as having file size limits, as batch processing, as local processing, as including format conversion, or as gated by an account or plan.
    What it says now
    The job keeps merge and gains the scope merging documents into one single combined PDF in the DocHub web editor, from the Page Manager or Page Controls panel using the merge and append file option, importing one or more PDFs from the device, cloud storage or the users own DocHub account and choosing whether the incoming file is inserted before or after a page of the current document, with the combined document remaining editable and able to be sent for signatures, from https://help.dochub.com/knowledge-base/document-editing/merging-documents-within-dochub and https://www.dochub.com/blog/how-to-merge-pdf-online/ read on 2026-10-16. The help article, found through the help centre document editing index, says DocHub offers the ability to merge documents together creating one single combined document, useful for sending several documents or templates as a package, that all functionality remains after merging so the combined document can be edited and sent for signatures, and that the flow starts in Page Manager where a modal offers to import a file or pick a PDF from Your DocHub. The blog walks the same flow: Page Controls, Add pages, the Merge and Append File menu, files from device, cloud or Your DocHub, one or more PDFs, insertion before or after a particular page, Insert, then download, export or share from a browser with nothing installed. The scope claims no bookmark preservation, no lossless behaviour, no file size limits, no batch processing, no local processing, no conversion and no account or plan requirement, since none of those is stated for the merge operation. The blog does mention converting images or other file types to PDF around the workflow, but that is adjacent guidance and stays with the separate convert job. Page reordering, rotation and deletion after merging are page management rather than merge and are outside the scope. The settled edit and sign scopes, pricing, privacy and platform findings and every other attribute are unchanged.

  165. 7a053ac

    The DocHub record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as an encryption algorithm claim, as print or copy permissions, as key management, as local processing, as a plan requirement, or as secure link and account access rather than protection of the PDF itself.
    What it says now
    The job keeps protect and gains the scope adding password protection to a downloaded PDF with the password protect option in the Encrypt group of the PDF Options in DocHub download or export dialog, which according to the help article adds password encryption to the downloaded PDF, so the protection is applied to the exported file rather than to the copy stored in the account, from https://help.dochub.com/knowledge-base/creating-importing-downloading-and-exporting-documents/download-or-export-your-pdf-documents read on 2026-10-16. The route mattered: three help centre category indexes were checked first, document editing, document permissions and sharing, and security privacy compliance and data retention, and none carries a dedicated protection article, so the operation was found where it is actually documented, in the PDF Options section of the download and export article. That section groups options under Flatten, Certify, Encrypt and Compress, and under Encrypt the single option is Password protect, described as add password encryption to your downloaded PDF. The scope claims no encryption algorithm since only password encryption is named, no key management, no print copy or editing permissions, no local processing and no plan or account requirement. It says protection applies to the exported file because the control lives in the export dialog, which keeps it from being read as a persistent property of the stored document or confused with secure links and account sharing, which are separate mechanisms in other help articles and were not folded in. Flatten, Certify and Compress sit beside Encrypt and are not claimed here; certify is already covered by the settled sign scope. The settled edit, sign and merge scopes, pricing, privacy and platform findings and every other attribute are unchanged.

  166. 7a053ac

    The CommonLook Office record, in the primary accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as including Excel or other Office formats, as remediating an existing PDF, as an OCR or batch tool, as legal compliance or automatic correctness, or as generic Office authoring.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope generating accessible PDFs from Microsoft Word (doc, docx) and PowerPoint (ppt, pptx) documents with CommonLook Office, which guides the author through accessibility checkpoints before the PDF is generated, maps Word styles to PDF tags, tags simple complex and presentation-style tables, and verifies the result against Section 508, HHS, WCAG or PDF/UA, from https://allyant.com/pdf-accessibility-software-solutions/commonlook-office/ read on 2026-10-16. The page describes CommonLook Office as a solution for creating accessible PDF documents that can generate compliant PDF files from Word and PowerPoint documents, surpassing the native accessible PDF functionality in Office; it says the software identifies and assists in resolving accessibility issues and automates most of the necessary tasks, guides users through Section 508, HHS, WCAG and PDF/UA, and lets users input accessibility information before the PDF is generated. Its feature list names step by step checkpoints, compliance against those four standards, complete verification of conformance criteria, style mapping to PDF tags, and table accessibility for simple, complex and presentation-style tables. The record needs the product distinction stated plainly: CommonLook PDF remediates PDFs that already exist, while CommonLook Office works from Word and PowerPoint sources and produces the tagged PDF, which is why this scope is about generating the PDF rather than repairing one. No legal compliance and no correctness guarantee is claimed, since the compliant wording is a vendor claim about outcomes, and the scope names only the standards verified against. Excel and every other Office format are excluded because the page names only Word and PowerPoint. No OCR, batch behaviour, AI accuracy or generic Office authoring is claimed. The settled attributes, privacy, pricing and platform findings and every other field are unchanged.

  167. 7a053ac

    The Docling record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as claiming every format the project lists, as an accuracy or handwriting claim, as OCR or audio and video parsing, as cloud processing, or as a general conversion or editing tool.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting structured content from documents with the Docling Python library or CLI, which parses formats including PDF, DOCX, PPTX, XLSX, HTML, EPUB and images into the unified DoclingDocument representation and exports it to Markdown, HTML, JSON, DocTags or DocLang, with the advanced PDF understanding the project lists as page layout, reading order, table structure, code and formulas, from https://raw.githubusercontent.com/docling-project/docling/main/README.md read on 2026-10-16. The README describes Docling as parsing diverse formats including advanced PDF understanding, lists parsing of PDF, DOCX, PPTX, XLSX, HTML, EPUB, Apple formats, email formats, images, LaTeX and more, lists its advanced PDF understanding as page layout, reading order, table structure, code, formulas and image classification, names the unified DoclingDocument representation, and lists export formats including Markdown, HTML, WebVTT, DocLang, DocTags and lossless JSON. Its quickstart shows the CLI writing a markdown file with structured content and the Python path converting a source and calling export_to_markdown. The scope claims no accuracy, no handwriting quality, no universal format coverage since the formats named are a subset of the longer published list, no arbitrary PDF editing and no redaction. Cloud processing is not claimed because the source describes local execution capabilities and the API server is a deployment option rather than a processing location statement. OCR, the visual language models, and the audio and video parsing are treated as separate capabilities and left out, and the export formats are described as output shapes rather than a general conversion claim. The open source, MIT licence, self hostable and locality attributes, pricing, privacy and platform findings and every other field are unchanged.

  168. 7a053ac

    The Docling record, in the convert job, scope decision and catalog semantics

    What it said
    The convert job carried no scope. A first pass recorded it as deliberately unresolved on the view that Docling merely serialises its own representation, which understated the record: the library exposes a distinct documented export surface, and the catalog already models exactly that shape of capability as a secondary convert job elsewhere.
    What it says now
    The job keeps convert as a secondary job and gains the scope renditions only: exporting the parsed Docling document to serialized files with the library export methods, Markdown through export_to_markdown, HTML through export_to_html, plain text through export_to_text, DocTags and DocLang, VTT and a dictionary form, which serialise the Docling Document rather than converting between document formats, with no DOCX, PPTX, XLSX or PDF output, no round trip and no fidelity claim. Sources read on 2026-10-16 are https://docling-project.github.io/docling/usage/supported_formats/ and https://docling-project.github.io/docling/reference/docling_document/. The supported formats page says Docling can parse various documents formats into a unified representation, the Docling Document, which it can export to different formats too, and gives separate input and output tables. Inputs include PDF, the Office formats, EPUB, HTML, CSV, images, audio, video and email. Outputs include HTML, Markdown, JSON described as lossless serialization of Docling Document, DocLang XML, text, DocTags, WebVTT, DocLang archive, chunks as JSONL and LaTeX. The Docling Document API reference lists the export methods by name including export_to_markdown, export_to_html, export_to_text, export_to_doctags, export_to_doclang, export_to_vtt and export_to_dict. The deciding evidence was catalog consistency rather than the Docling docs alone: adobe-pdf-extract-api carries a secondary convert job scoped as renditions only, structured JSON or Markdown output with tables optionally as CSV or XLSX and images as PNG, not general PDF conversion, which is the same situation, an extraction product whose convert job covers serialised output rather than document format conversion. The scope therefore says renditions only and excludes generic office and PDF round trips, DOCX, PPTX, XLSX and PDF output, fidelity and round trip claims, PDF generation, OCR and universal format coverage. The settled extract scope, the MIT and open source and local and self hostable attributes, pricing, privacy and platform findings are unchanged. This entry supersedes an earlier uncommitted draft that argued the other way; that reasoning is recorded so the reversal is visible rather than silent.

  169. 7a053ac

    The Docling record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as an accuracy or handwriting claim, as universal language coverage, as automatic engine quality ranking, as searchable PDF output, or as implying that OCR is a cloud service or needs a GPU.
    What it says now
    The job keeps ocr as a secondary job and gains the scope running OCR on scanned pages and images in Docling with a selectable engine among RapidOCR, Nemotron-OCR, EasyOCR, ocrmac, tesseract CLI and tesserocr, each installed as an extra package, with the language set through the single OcrOptions.lang field either as engine native codes or as iso: prefixed BCP-47 tags, an error raised when the selected engine has no model for the requested language, and an option to send OCR requests to a remote KServe v2 inference server, from https://docling-project.github.io/docling/concepts/OCR/ read on 2026-10-16, with the README line on extensive OCR support for scanned PDFs and images as corroboration. The page lists the six engines as installable extras and the KServe v2 remote inference option; it says every engine takes languages through OcrOptions.lang in exactly two forms, the engine own code handed over untouched or an ISO prefixed BCP-47 tag canonicalised and mapped onto the engine notation, with supported_ocr_languages reporting what each engine serves; and it states that an exception is raised when an engine cannot serve the input language and that Docling never quietly substitutes a different recognizer. It also documents that RapidOCR and Nemotron-OCR run one language per conversion and take the first tag while warning about the rest, that the KServe client canonicalises nothing, and that RapidOCR model size chooses between tiny, small and medium checkpoints for PP-OCRv6 with the project explicitly not benchmarking or recommending one. The scope claims no accuracy, no handwriting, no universal language coverage, no automatic engine quality ranking, no searchable PDF generation, no PDF/A, no redaction and no GPU requirement, and it does not claim general cloud processing because the KServe option is documented as a remote inference client choice. The settled extract and convert scopes, the local, MIT, open source and self hostable attributes, pricing, privacy and platform findings are unchanged.

  170. 7a053ac

    The Documenso record, in the primary sign job

    What it said
    The sign job carried no scope, so it could be read as signer certificate signing, as a PAdES or qualified signature claim, as identity assurance, as a legal enforceability or compliance guarantee, as OCR or redaction or PDF content editing, or as a cloud only product.
    What it says now
    The job keeps sign, remains primary, and gains the scope running an electronic signature workflow in Documenso by uploading a PDF or starting from a template, assigning recipients as signers, approvers, viewers, assistants or CC, placing signature and other fields on the document and sending it for signing in parallel or in sequence, after which each recipient signs by drawing, typing or uploading a signature image and the completed PDF is sealed at the platform level with an X.509 certificate that proves integrity and authenticity, with an audit log attached and an optional trusted timestamp from a time stamping authority, from https://docs.documenso.com/docs/concepts/document-lifecycle, /recipient-roles, /field-types and /signing-certificates read on 2026-10-16, with the repository README for licence and self hosting context. The lifecycle page gives the draft, pending, completed and rejected states, the draft actions of adding recipients and roles, fields and signing order, sending to pending, the completion conditions, and the sealed certificate, attached audit log and unmodifiable completed document. The roles page lists signer, approver, viewer, assistant and CC with whether each must act and whether each can sign, plus parallel default and sequential order. The field types page documents the signature field as drawn, typed or uploaded and the other field types, with each signer requiring at least one signature field. The signing certificates page states the completed document is digitally signed using X.509 certificates, describes hashing the document and signing the hash with the certificate private key, states the signature is applied at the platform level rather than by individual signers, names integrity and authenticity as the guarantees, and documents optional trusted timestamps from a time stamping authority plus CA issued versus self signed trust results. The scope claims no PAdES profile, no qualified signature level, no identity assurance, no legal enforceability and no compliance guarantee, and no OCR, redaction, PDF content editing or generic form building. It names neither cloud nor self hosting as the only deployment, so the settled platforms and attributes are unchanged.

  171. 7a053ac

    The DocuSeal record, in the primary sign job

    What it said
    The sign job carried no scope, so it could be read as signer certificate signing, as PAdES or a qualified signature level, as identity assurance, as a legal enforceability or compliance guarantee, as OCR or redaction or PDF content editing, or as a cloud only product.
    What it says now
    The job keeps sign, remains primary, and gains the scope running a document signing workflow in DocuSeal by uploading a document or starting from a template and placing fields with the PDF form builder, adding multiple submitters who are invited by automated email to review, fill and sign, with signatures drawn or typed in the signing tool or applied as a signature image, after which each submitter receives a copy of the filled and signed files and the completed PDF is automatically signed with a platform PDF eSignature that can be verified through the platform for authenticity, with an audit log available alongside the completed document, from https://www.docuseal.com/docs/welcome-to-docuseal, /guides, /guides/download-signed-documents, /online-signature and /sign-documents-online read on 2026-10-16, plus the repository README. The welcome page documents the form builder and field types, multiple submitters each receiving their own copy of filled and signed files, automated SMTP emails carrying the invitation or a copy, and an automatic eSignature feature stating that all documents are automatically signed with a digital industry standard PDF eSignature and that signed files can be verified through the platform for authenticity. The guides index describes retrieving the completed PDF and its audit trail, and the download guide shows the form.completed webhook carrying per submitter documents with timestamps and a submission audit_log_url and combined_document_url. The signature generator page documents drawing or typing a signature and applying a transparent PNG image. The sign documents page lists supported upload formats with a 100MB maximum. The scope keeps the submitter signature and the platform applied automatic eSignature distinct, claims no PAdES, no certificate authority trust, no qualified signature, no identity assurance, no legal enforceability and no compliance guarantee, and does not fold in the separately marketed ESIGN, eIDAS levels, qualified signature or knowledge based authentication pages. No OCR, redaction or PDF content editing is claimed. Neither cloud nor self hosting is named as the only deployment, so the settled platforms and attributes are unchanged.

  172. 7a053ac

    The Docusign record, in the primary sign job

    What it said
    The sign job carried no scope, so it could be read as including Advanced Electronic Signatures or Qualified Electronic Signatures, as signer certificate or identity verification claims, as eIDAS compliance or legal enforceability, as notarization, as audit or certificate of completion evidence, or as OCR or PDF editing.
    What it says now
    The job keeps sign, remains primary, and gains the scope running an electronic signature workflow in Docusign by creating an envelope from documents or a template, assigning each recipient a recipient type that defines their role, including signers, in person signers, carbon copy recipients, certified delivery recipients, agents, editors and intermediaries, placing tabs that give each recipient input such as a Sign Here tab positioned by fixed, AutoPlace or PDF form field transformation, and sending the envelope to remote recipients by email or to embedded recipients inside an application, with signing complete once recipients have entered data for all of their tabs. Sources read on 2026-10-16 through the repository headless renderer, because the Docusign developer documentation is client rendered, are the recipients, tabs and standards based signatures concept pages under https://developers.docusign.com/docs/esign-rest-api/esign101/concepts/. The recipients page describes envelopes, one or more recipients each with an assigned recipient type, one or more tabs per recipient, remote versus embedded recipients, and a recipient type table covering agents, carbon copies, certified deliveries, editors, in person signers, intermediaries and notary recipients with routing order semantics. The tabs page describes tabs as places a recipient must provide input, Sign Here requiring a signature and Text prompting text, the three positioning methods of fixed positioning, AutoPlace anchor tagging and PDF form field transformation, and completion as the point when recipients have entered data for all of their tabs. The standards based signatures page separates ordinary electronic signatures from Advanced Electronic Signature using certificates and from Qualified Electronic Signature from Qualified Trust Service Providers, both of which need signature providers via recipientSignatureProviders, with the qualified path tied to identity verification and the eIDAS regulation. The scope therefore claims no AES, no QES, no signature provider, no PAdES, no identity verification, no eIDAS compliance and no legal enforceability, and it leaves out the specialised notary recipient type and claims no certificate of completion or audit trail evidence because those pages were not read in this pass. No OCR, redaction or PDF editing is claimed and no data residency claim is added beyond the existing record attribute.

  173. 7a053ac

    The Dompdf record, in the primary convert job

    What it said
    The convert job carried no scope, so it could be read as a browser based renderer with modern CSS support, as a fidelity guarantee, as supporting PDF editing or Office round trips or PDF/A or PDF/X, or as a hosted service.
    What it says now
    The job keeps convert, remains primary, and gains the scope converting HTML and CSS to PDF with the Dompdf PHP library, which renders HTML passed to loadHtml together with inline styles, linked stylesheets and presentational HTML attributes using a layout engine the project describes as mostly CSS 2.1 compliant with a few CSS3 properties, produces the PDF through render and emits it through stream, and lets the caller set paper size and orientation through setPaper, with the project documentation stating that CSS flexbox and CSS Grid are not supported, from https://raw.githubusercontent.com/dompdf/dompdf/master/README.md read on 2026-10-16. The README describes Dompdf as an HTML to PDF converter and as mostly a CSS 2.1 compliant HTML layout and rendering engine written in PHP that reads external stylesheets, inline style tags and element style attributes and supports most presentational HTML attributes; its features include most CSS 2.1 with a few CSS3 properties, import media and page rules, local and http or ftp stylesheets, complex tables, gif png bmp and jpeg images and inline PHP. The quick start shows new Dompdf, loadHtml, optional setPaper, render and stream. The limitations section states that CSS flexbox and CSS Grid are not supported, that table cells are not pageable, that raw inline SVG does not work yet and that one instance should not render more than one document. The scope claims no fidelity or pixel perfect guarantee, no PDF editing, no Office round trips, no PDF/A or PDF/X, no OCR, no browser automation since the renderer is the library own engine rather than a browser, and no cloud processing because this is a PHP library that runs where installed, consistent with the record locality attribute. The settled attributes, pricing and platform findings are unchanged.

  174. 7a053ac

    The Dropbox Sign record, in the primary sign job

    What it said
    The sign job carried no scope, so it could be read as signer certificate signing, as PAdES or qualified signatures, as an identity verification claim through the electronic ID option, as a legal enforceability or compliance guarantee, as audit or certificate evidence, or as OCR or PDF editing.
    What it says now
    The job keeps sign, remains primary, and gains the scope creating and sending a signature request through the Dropbox Sign API from uploaded or URL referenced documents, delivered to signers by email, with signers added by name and email address and an optional order, grouped signers, CC recipients, fields placed per document as signature, initials, date signed, text, checkbox, radio, dropdown or hyperlink fields or through text tags, and signature options allowing a signature to be drawn, typed or uploaded, where the API returns a signature request id, a signing URL and per signer status with signed timestamps and the completed files are available through the files endpoints, from https://developers.hellosign.com/api/signature-request/send.md and https://developers.hellosign.com/docs/overview.md read on 2026-10-16. The send reference says the endpoint creates and sends a new SignatureRequest with the submitted documents and that if form fields are not specified a signature page is affixed for all signers, documents the signer object with name, email address, optional order, pin and SMS phone, grouped signers, CC addresses, subject message and title, text tags, form field types of text, dropdown, hyperlink, checkbox, radio, signature, date signed and initials plus merge variants, and a signing_options object with draw, type, upload and phone flags and a default type. Its response carries signature_request_id, is_complete, is_declined, files_url, signing_url and a signatures array with status_code, signed_at, last_viewed_at and last_reminded_at. The overview page describes the API as eSignature tools, says test_mode requests are not legally binding, and lists which endpoints each plan unlocks. The scope claims no signer held certificate, no PAdES, no qualified signature, no identity verification, no legal enforceability and no compliance guarantee, and it leaves out the is_eid electronic identification add-on which needs an identity verification provider. The legally binding wording in the overview is recorded here rather than turned into a scope claim, and no audit trail or certificate of completion evidence is claimed because those pages were not read. No OCR, redaction or PDF editing is claimed and no data residency or deployment claim is added beyond the existing record attributes.

  175. 7a053ac

    The Equidox record, in the primary accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as a compliance guarantee, as an AI accuracy claim, as OCR, as generic PDF editing or conversion, or as the high volume batch product. The record website URL was checked as part of this pass and already pointed at the vendor site, so it needed no change.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope remediating an existing PDF for accessibility with Equidox Software, using the AI Smart Zone Detector to detect and tag text and headings, the List Detector for lists including nested lists and the Table Detector for rows, columns and complex cell structures, correcting reading order with the one-click re-order tool, adding image alt text or marking decorative images as artifacts, entering form field tooltips in the Form Pane, and checking the structure in an HTML preview before export, with the software available as cloud SaaS for PC or Mac or as an on-premises web appliance, from https://equidox.co/pdf-solutions/pdf-accessibility-software/ read on 2026-10-16. That page documents each operation by name: the Smart Zone Detector detecting elements even in PDFs without existing tags, the one click Re-Order tool for reading order with an HTML preview, the List Detector for multi level and nested lists, the Smart Table Detector for rows columns and complex cell structures with multi level headers and spanned cells, the Form Pane for tooltips with tagging applied at export, alt text through an Image Pane with decorative images artifacted, heading shortcuts with validation before export, and Zone Transfer copying a tag structure between PDFs. Deployment is described as cloud based SaaS for PC or Mac or an on premises web appliance. The scope claims no legal compliance despite the WCAG, Section 508, ADA, AODA and EAA marketing, no AI accuracy, no OCR, no generic editing, conversion, compression or redaction, and no batch behaviour, which belongs to the separate Equidox AI high volume offering and was left out. The record website URL already pointed at https://equidox.co/, which is the vendor site, so no field besides the scope was changed; the .com hosts redirect cross-origin to gcd.com and were treated as non vendor content rather than followed. The settled attributes, pricing and platform findings are otherwise unchanged.

  176. 7a053ac

    The Equidox ledger entry itself, correcting a claim in the preceding entry and in one commit message

    What it said
    The Equidox accessibility ledger entry, and the message of the content commit it cites, both said that the record website URL had been corrected from the .com domain to https://equidox.co/. That was wrong. The record already pointed at https://equidox.co/ before the pass, the content commit changed only the accessibility scope, and the script that would have corrected the URL reported that it made no change. The claim came from an assumption that the field matched the domain that was being opened, not from reading the field first.
    What it says now
    Corrected in the entry text and recorded here. The record website URL was already https://equidox.co/, which is the vendor site, so no field besides the accessibility scope was changed in this pass. What is true and worth keeping is the finding that equidox.com and www.equidox.com both redirect cross-origin to gcd.com, which is not vendor content, so those hosts should not be treated as Equidox documentation. The scope added in the same pass is unaffected by this correction and stands: remediating an existing PDF with Equidox Software, the Smart Zone Detector, List Detector and Table Detector, reading order correction, image alt text and decorative artifacts, form field tooltips and an HTML preview before export, as cloud SaaS or an on premises appliance. The commit message of 7d92921 still contains the wrong sentence and cannot be rewritten without amending a cited commit, so this entry is the correction of record.

  177. 7a053ac

    The Documenso record, in the edit job, scope exclusion

    What it said
    The edit job has no scope, so it could be read as content editing of an existing PDF, meaning changing text, images, layout, annotations or pages. The presence of a document editor and an Edit a Template section makes that reading plausible to someone who has not seen what the editor contains.
    What it says now
    Left unresolved deliberately, because the documentation describes preparation and workflow configuration rather than content editing, and the user guide has no page describing a change to PDF content at all. Pages read on 2026-10-16 are https://docs.documenso.com/docs/users (the user guide index), https://docs.documenso.com/docs/users/documents/upload and https://docs.documenso.com/docs/users/templates/create. The upload page says Documenso accepts PDF and DOCX, that a PDF is used as-is while a DOCX is converted to PDF on the server during upload and the converted PDF is what recipients sign, that encrypted and password-protected PDFs are not supported, that the maximum file size is 50MB, and that after upload the editor opens so the user can add recipients and fields. So the editor is described as the place to add recipients and fields, not to change document content. The templates page is where the word edit appears, under Edit a Template, and it lists the things that can be changed as recipients, fields, or settings, adding that changes only affect future documents created from the template. Nothing in these pages documents changing existing text, images, layout, annotations or page content. Counting this as edit would duplicate the settled sign scope, which already covers uploading, recipients and field placement, and it would overstate the product as a PDF editor. Two adjacent capabilities are noted here so they are not confused with edit: combining several uploaded PDFs into one envelope or template is envelope assembly rather than page editing, and the DOCX to PDF conversion on upload is a conversion, which is not editing and which this record has no convert job for. No OCR and no redaction is documented on these pages either.

  178. 7a053ac

    The Foxit PDF Editor record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as covering desktop and web content editing, or as covering OCR, redaction, conversion, signing or accessibility remediation, none of which was evidenced for this job in this pass.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing existing PDF content in the Foxit PDF Editor mobile edition, where the official user manual for version 11.2.0 documents editing an existing text object by placing the cursor in the text box and typing or removing text and changing the font, size, color and alignment, moving or deleting a text paragraph, and moving, resizing, rotating, replacing, exporting or deleting an existing image object, described in the manual as functionality available to subscribers, from https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Edit_PDF_Files.html read on 2026-10-16. The manual page lists, for subscribers, adding and editing text objects, adding and editing image objects, inserting audio and video, creating a hyperlink and undo and redo; it describes text editing as tapping the text box, choosing Edit, placing the cursor and typing or removing text then changing font size colour and alignment, and image editing as moving, resizing, rotating, replacing, exporting or deleting the image. The product page at https://www.foxit.com/pdf-editor/, rendered for this pass, describes the product as creating, editing, converting, signing and managing PDFs across desktop, mobile and web, and its edition comparison places desktop and web editing in Foxit PDF Editor while the mobile apps are added in Foxit PDF Editor+. The surface is named because the desktop and web manuals were not readable in this pass: the Windows manual path returned HTTP 500 and the context sensitive Edit Objects help page also returned HTTP 500, so no desktop or web operation is claimed and the product level statements are recorded as marketing rather than specific capability. The scope claims no OCR, redaction, conversion, compression, merge, signing, accessibility remediation, fidelity, universal format support, AI accuracy or cloud processing, and it leaves out the documented additions of new text objects, audio, video and hyperlinks because they add rather than edit existing content. The settled compress and compare scopes, attributes, pricing and platform findings are unchanged.

  179. 7a053ac

    The Foxit PDF Editor record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as including an encryption algorithm claim, key management, DRM beyond what is documented, certificate trust, signing, PDF/A or a legal confidentiality guarantee, and as applying to desktop and web rather than to the surface actually read.
    What it says now
    The job keeps protect and gains the scope protecting PDF files in the Foxit PDF Editor mobile edition, where the official user manual for version 11.2.0 documents setting an open document password and switching on document restrictions that disable selected actions, with the same password required to remove security, protecting a file with a certificate on Android only using a certificate stored on the device, redacting text, images and pages, including by search, with the redaction applied permanently and not undoable, and encrypting with Microsoft Information Protection rights policy templates, with files protected that way saved as PPDF files, from https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Protect_PDF_Files.html read on 2026-10-16. The page lists four protection methods for subscribers. Password and restrictions: Protect then File Encryption, Open Document toggle with a password, Add Document Restrictions toggle with individual action toggles, and a note that the password is needed to open the file and to remove the restricted permissions. Remove Security: Protect then Remove Security, password required to remove restrictions. Certificate protection: Protect then Certificate Protect, select a certificate and enter its password, stated as Android only, Android 2.3 and above, with the digital certificate saved on the device first. Redaction: Protect then Redaction, with redact text or image, redact pages and search and redact, appearance settings for colour text font and size, Apply and Apply all, and a tip that applied content is permanently redacted and removed, cannot be undone and that saving a copy first is recommended. Microsoft Information Protection: Protect then Microsoft Information Protect, sign in, choose a rights policy template, optionally custom permissions with read change and full control, expiry, print copy and comment toggles, a permission check on open and a security watermark, with a note that PDF documents protected this way are saved as PPDF files. The scope claims no encryption algorithm, key management, DRM beyond that named product, certificate trust, PDF/A, signing or legal confidentiality guarantee. The surface is the mobile edition because that is the manual read, and as in the previous pass on this record the desktop and web manuals were not readable, so no desktop or web protection claim is made. The settled edit, compress and compare scopes, attributes, pricing and platform findings are unchanged.

  180. 7a053ac

    The Foxit PDF Editor record, in the ocr job, retrieval result and blocker

    What it said
    The ocr job has no scope. The product supports OCR according to its own comparison table, so the risk is that a later pass scopes the job from that one word and lets a reader assume searchable output, language support, accuracy or desktop behaviour that no readable page states.
    What it says now
    Left unresolved, and the reason is coverage rather than absence of the feature. What was read on 2026-10-16: the mobile user manual index at https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/index.html lists twenty two pages and none of them is an OCR page; the only scanning page in that manual, https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Scan_Documents.html, was read in full and describes scanning paper documents to PDF, adding pages, deleting pages, rotating and cropping pages, adjusting colour brightness and contrast, and saving or sharing the scan, with no mention of text recognition, OCR or searchable text anywhere on it, and several guessed OCR page names in the same directory returned 404. The product page at https://www.foxit.com/pdf-editor/, rendered in a previous pass, lists OCR only as a yes or no capability row, OCR scanned documents, marked Yes for PDF Editor+ and Adobe Acrobat Pro, which names no operation, no input, no output format, no language and no limit. The desktop and web manual pages could not be read: in the previous pass the Windows manual path returned HTTP 500 and the context sensitive Edit Objects help page also returned HTTP 500, and no desktop OCR page has been read. A marketing line elsewhere on the site about making scanned documents searchable and editable is not documentation of an operation. With no readable page describing the OCR operation, its input types, its language or mode limits or its output behaviour, the scope is not written, because the narrowest defensible scope cannot be derived from a capability check box. The blocker is a documentation surface that will not load, and the path to unblock it is the desktop user manual for a version that renders, or a Foxit help article on recognising text.

  181. 7a053ac

    The Foxit PDF Editor record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format support, as including OCR or merge, as covering blank or template creation, as a PDF/A conversion, as a fidelity or accuracy guarantee, or as a desktop and web capability.
    What it says now
    The job keeps convert and gains the scope converting files to PDF and PDFs to other formats in the Foxit PDF Editor mobile edition, where the official user manual for version 11.2.0 documents creating a PDF from Microsoft Word, Excel, PPT, TXT, HTML and image files and from a webpage URL, and exporting a PDF to Microsoft Word in doc or docx format, Excel, PowerPoint, text as plain or rich text, an image or HTML, with the converted file saved in the same folder as the source PDF file, from https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Create_and_Export_PDF_Files.html read on 2026-10-16. The page says the app allows subscribers to create PDFs and export PDFs to other file formats; it documents File to PDF for Word, Excel, PPT, TXT, HTML and image files and Web to PDF for a URL with a Convert action; and it documents Export PDF to Microsoft Word in doc or docx, Excel, PowerPoint, text as plain or rich text, image or HTML, with the converted file saved in the same folder as the source PDF. The formats are named exactly, so no universal coverage is claimed. No OCR is claimed, since this page recognises no text and the ocr job is separately blocked on an unreadable desktop manual. Merge is documented on the same page but is left to its own job so it is not double counted, and blank creation and template creation are excluded as creation rather than conversion, with the template route also marked English Edition only. No PDF/A, compression, fidelity, accuracy or cloud processing is claimed, and no browser automation is claimed for Web to PDF beyond the documented URL and Convert steps. The surface is the mobile edition because that is the manual read, and the desktop and web manuals have returned server errors in earlier passes. The settled edit, protect, compress and compare scopes, attributes, pricing and platform findings are unchanged.

  182. 7a053ac

    The Foxit PDF Editor record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as offering control over the order the files are combined in, as preserving bookmarks or metadata, as lossless, as supporting formats other than PDF, or as including page organisation and split.
    What it says now
    The job keeps merge and gains the scope merging multiple PDF files into a single PDF in the Foxit PDF Editor mobile edition, where the official user manual for version 11.2.0 documents choosing Merge PDFs, selecting the documents to combine and confirming the selection, optionally deleting an unwanted file before merging and optionally renaming the output, after which the app reports where the merged PDF was saved, from https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Create_and_Export_PDF_Files.html read on 2026-10-16. That page Merge PDF Files section says you can combine and merge multiple PDF files into a single one, and gives the steps as choosing Merge PDFs, selecting the documents and tapping Done, with Edit available to delete an unwanted one, tapping the title to rename the file, and confirming, after which a message reports where the PDF is saved with Open Folder and Open File actions. Ordering is deliberately not claimed: a search of the whole page for order, sort and arrange returned no occurrences, so the scope describes selection and removal only and a reader is not told there is a sequence control. Preservation of bookmarks or metadata, losslessness and non PDF inputs are not claimed either, and the page contains no bookmark, metadata or lossless wording. Page organisation after merging, which the same page mentions with a pointer to Organize PDF Pages, belongs to page editing and is left out, as are split, page insertion, conversion, compression and PDF/A. The surface is the mobile edition because that is the manual read, and the desktop and web manuals have returned server errors in earlier passes. The settled edit, protect, convert, compress and compare scopes, attributes, pricing and platform findings are unchanged.

  183. 7a053ac

    The Foxit PDF Editor record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as PAdES or qualified signature support, as legal effect or identity assurance, as certificate authority trust, as applying only to drawn signatures, or as including form fields, editing or protection.
    What it says now
    The job keeps sign and gains the scope signing a PDF in the Foxit PDF Editor mobile edition, where the official user manual for version 11.2.0 documents creating a signature by drawing it, inserting an image of it from the device or capturing it with the camera and then placing it on a page, or on a subscription creating a signature with a digital certificate stored on the device by selecting the certificate and entering its password, and verifying a digital signature already present in a PDF through the Digital Signatures pane with options to view the certificate and add it to a trusted certificates list, with an applied signature described as no longer editable or movable and earlier signed versions accessible from the same pane, from https://help.foxit.com/manuals/pdf-editor/mobile/en_us/11.2.0/Add_Signatures.html read on 2026-10-16. The page documents creating a signature by drawing with colour and thickness, inserting an image from the device or capturing with the camera, and a subscription required path to create a signature with a digital certificate by selecting a certificate and entering its password with the certificate saved on the device first. It documents placing a signature on the page and applying it with the Sign or Add Signature action, with a tip that the applied signature cannot be edited or moved and that the signature and other changes will be saved; verifying a signature through the Digital Signatures pane with results on the document pane and options to view the certificate and trust it, adding certificates to a trusted certificates list under Protect, and viewing earlier signed versions from the same pane. The scope keeps the drawn or image signature and the certificate based signature distinct and says which needs a subscription. It claims no PAdES profile, no eIDAS or qualified signature, no legal effect, no identity assurance, no certificate authority trust guarantee and no timestamp or long term validation, and the trusted list is described only as the documented operation. Form fields, editing, protection and redaction are separate jobs and are not folded in. The surface is the mobile edition because that is the manual read, and the desktop and web manuals returned server errors in earlier passes. The settled edit, protect, convert, merge, compress and compare scopes, attributes, pricing and platform findings are unchanged.

  184. 7a053ac

    The Foxit PDF SDK record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as covering OCR, redaction, conversion, merge, signing or accessibility tagging, as applying to the mobile or web SDKs, as available without the AdvEdit licence or on older SDK versions, or as a fidelity or format claim.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing existing PDF content with the Foxit PDF SDK for Desktop and Server, where the developer guide documents page objects for adding and removing text, path, image and form objects on a page and setting their properties, and a paragraph editing module that inserts, deletes and modifies existing text including its content and format, with join, split, link and unlink operations on text blocks, supported on Windows, Linux and Mac from C, C++, Java, C#, Python, Objective-C and Go, requiring the AdvEdit add-on licence and SDK version 10.0 or later, from https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/page-management/page-object.html and https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/content-and-search/paragraph-editing.html read on 2026-10-16. The page object page says page objects let developers work with text, path, image and form objects, that the SDK provides APIs to add and remove PDF objects on pages and set their properties, and lists headers and footers, image logos and templates as uses, with samples that create a text object, set text and text state, insert the graphics object, transform it and generate content. The paragraph editing page says the module provides adjustment, merging and splitting functions, with ParagraphEditing operations of insert text, delete text, modify text including content and format, and format adjustment, and JoinSplit operations of join, split, link and unlink, and it states its own boundaries as platform Windows Linux Mac, languages C C++ Java C# Python Objective-C Go, licence requiring the AdvEdit module, and SDK version 10.0 or later. The scope claims no OCR, redaction, conversion, merge, signing, protection or accessibility tagging, each documented on its own page of the same guide and belonging to other jobs or to no job here, no viewer behaviour and no fidelity or format claim. It names the Desktop and Server SDK because the quick guide documents mobile and web SDKs separately. The settled convert scope, attributes, pricing and platform findings are unchanged.

  185. 7a053ac

    The Foxit PDF SDK record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a drawn signature feature, as identity assurance or a qualified signature, as eIDAS or legal effect, as a guarantee that signatures validate, or as covering the mobile and web SDKs.
    What it says now
    The job keeps sign and gains the scope creating, verifying and removing digital signatures in PDF documents with the Foxit PDF SDK for Desktop and Server, using the built-in signature callbacks for the Adobe.PPKLite filter with the adbe.pkcs7.detached subfilter and a SHA-256 digest or the legacy adbe.pkcs7.sha1 subfilter, signing with a certificate file and its password, getting and setting signature properties and customising signature field appearance, and creating PAdES signatures with the ETSI.CAdES.detached subfilter at the B-B, B-T, B-LT and B-LTA levels the guide defines, with timestamp servers configured through the TimeStampServer and TimeStampServerMgr classes, PAdES being documented as supported from SDK version 7.0, from https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/signatures/signature.html and https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/signatures/PAdES.html read on 2026-10-16. The Digital Signatures page lists create, verify, remove, get and set properties, display signatures and customise signature field appearance, documents the Adobe.PPKLite filter with adbe.pkcs7.detached and adbe.pkcs7.sha1 subfilters, calls detached with SHA-256 the preferred path, and shows signing with a certificate file and password with a demo-certificate warning. The PAdES page says PAdES support starts at SDK version 7.0, applies CAdES signatures with default subfilter ETSI.CAdES.detached, defines the B-B, B-T, B-LT and B-LTA levels, documents the default callback and the TimeStampServerMgr and TimeStampServer classes, and documents methods to obtain a signature PAdES level. The scope describes cryptographic certificate signing and does not claim drawn signing, because the developer guide documents no drawn signature operation here. It claims no identity assurance, no qualified signature, no eIDAS or legal effect and no guarantee that a signature validates against a trust chain, since the page only advises enabling full trust-chain validation and SHA-256 or stronger digests for production. PAdES levels are named as documented definitions rather than a compliance assertion. Editing, protection, redaction, OCR, conversion and accessibility are separate chapters of the same guide and are not folded in, and no mobile or web SDK signing claim is made. The settled edit and convert scopes, attributes, pricing and platform findings are unchanged.

  186. 7a053ac

    The Foxit PDF SDK record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as an accuracy or handwriting claim, as PDF/A or compliance output, as available on macOS or mobile, as usable without the OCR module licence or its resource package, or as the same generic scanner behaviour the desktop product offers.
    What it says now
    The job keeps ocr and gains the scope running OCR on scanned documents and image-based or mixed PDFs with the Foxit PDF SDK for Desktop and Server, where the developer guide documents initialising an OCREngine with a resource directory and languages, configuring OCRConfig options for picture detection, denoise, deskew and resolution, running OCRPDFPage, OCRPDFDocument or OCRPDFDocuments to produce a searchable PDF that adds a text layer or an editable PDF that writes recognition results into document content, and using OCRConvertTo to export recognised content to Word, Excel, PowerPoint or HTML, on Windows and Linux x64 with a licence that includes the OCR module and an OCR add-on resource package, the ABBYY OCR resource package being required from version 12.0, and a command-line OCR tool documented from version 11.1 with page range, language, editable output and output format options, from https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/recognition/ocr.html and https://developers.foxit.com/documents/development-guide/pdf-sdk-desktop-server/features/recognition/ocr-tool.html read on 2026-10-16. The OCR page names the recognised inputs, the Windows and Linux x64 platform requirement, the OCR module licence and the resource package including the ABBYY package requirement from 12.0, the OCREngine OCR OCRConfig OCRCallback OCRProgressCallback and OCRSettingData classes, the integration flow and its configuration options, the two output types of searchable PDF and editable PDF, and the OCRConvertTo export to Office or HTML. The command-line page documents the tool from version 11.1 with its -type modes, input, output, engine, language, edit, password, range and format options and its exit codes. The scope claims no accuracy or handwriting, and the confidence setting is not promoted to a quality claim. No PDF/A or compliance, no macOS, mobile or web OCR, no generic scanner behaviour, and no text editing, conversion, redaction or accessibility beyond the OCRConvertTo export the OCR chapter documents are claimed. The settled edit, sign and convert scopes, attributes, pricing and platform findings are unchanged.

  187. 7a053ac

    The Genius Scan record, in the primary ocr job

    What it said
    The ocr job carried no scope, so it could be read as available on the free plan, as covering scanning and capture, as producing searchable PDF or PDF/A output, as a language breadth or accuracy claim, or as applying to the web app.
    What it says now
    The job keeps ocr, remains primary, and gains the scope applying text recognition to scans in the Genius Scan mobile app, where the help centre documents a Text Recognition setting that is switched on and then applied to new documents, a Manage Languages menu where at least one language must be active or the feature cannot process documents, and exporting the recognised text by copying it from the Text tab or by exporting the document as text, including several documents at once, with the help centre stating the feature requires an advanced plan and the pricing page listing OCR under Ultra rather than Basic, from https://help.geniusscan.com/document-management/text-recognition-ocr/enable-text-recognition.md, https://help.geniusscan.com/exporting-and-sharing/export-formats-and-options/export-scans-as-text.md and https://thegrizzlylabs.com/genius-scan/pricing read on 2026-10-16. The enabling page notes the feature requires an advanced plan and warns that at least one active language must be set in Manage Languages or Text Recognition cannot process documents, with iOS and Android steps that switch it on for new documents. The export page says a document with Text Recognition applied can be exported as Text, and documents copying or exporting text from the Text tab, Text on iOS and TXT on Android as the export format, and doing so for several documents at once. The pricing page lists OCR under Ultra rather than Basic and keeps PDF/A support as a separate row. The scope claims no accuracy or handwriting, no language breadth since no language list was read and the SDK list belongs to a different product, and no searchable PDF or PDF/A. Scanning and capture, conversion, editing and cloud processing are not claimed, and no web app claim is made because the help centre documents the steps for iOS and Android only. The settled merge and protect scopes, attributes and platform findings are unchanged.

  188. 7a053ac

    The gImageReader record, in the primary ocr job

    What it said
    The ocr job carried no scope, so it could be read as an OCR language coverage claim from the international language support bullet, as producing searchable PDF or PDF-A output, as an accuracy or handwriting claim, as cloud processing, or as including the hOCR to PDF step that belongs with conversion.
    What it says now
    The job keeps ocr, remains primary, and gains the scope recognising text in images and PDF documents with gImageReader, which the project describes as a Gtk and Qt front end to tesseract-ocr, importing images and PDFs from disk, scanning devices, the clipboard and screenshots, defining the recognition area manually or automatically, processing multiple images and documents in one go, and recognising to plain text or to hOCR with the recognised text displayed next to the image for post-processing including spell checking, from https://raw.githubusercontent.com/manisandro/gImageReader/master/README.md read on 2026-10-16. The README calls gImageReader a simple Gtk and Qt front end to tesseract-ocr and lists the import sources, batch processing, manual or automatic recognition area, recognition to plain text or hOCR, text displayed beside the image, post-processing with spellchecking, PDF generation from hOCR, and international language support. That language bullet links to Weblate and the desktop entry, so it concerns translating the application interface, and the scope therefore makes no OCR language coverage claim and asserts no language list. Generating PDF documents from hOCR is documented but left out, because it produces a PDF and the convert job on this record is already settled, so including it would double count. No searchable PDF, PDF/A, handwriting, accuracy, layout fidelity, redaction or cloud processing is claimed, and the only scanner mention is the documented import from scanning devices. The installation list covers Windows and Linux distributions, consistent with the record platforms. The settled convert and edit scopes, the GPL-3.0 licence, the open source and locality attributes and every other field are unchanged.

  189. 7a053ac

    The Gotenberg record, in the primary convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage, as including the screenshot routes, as covering the PDF engine manipulation routes such as merge, split, PDF/A or encryption, as a fidelity claim, as OCR or editing, or as a hosted service.
    What it says now
    The job keeps convert, remains primary, and gains the scope converting content to PDF over HTTP with Gotenberg, where the official routes overview states that every route accepts a multipart form data POST request and returns a file, and lists the conversion routes as converting a web page by URL at /forms/chromium/convert/url, an HTML file at /forms/chromium/convert/html, Markdown files at /forms/chromium/convert/markdown, and office documents such as docx, xlsx and pptx at /forms/libreoffice/convert, from https://gotenberg.dev/docs/getting-started/routes read on 2026-10-16. The page says every route accepts a multipart form data POST and returns a file, lists the four Convert to PDF routes with the Chromium and LibreOffice engines separated, and separately lists screenshot routes that return images, a Manipulate PDFs table covering merge, split, PDF/A and PDF/UA conversion, metadata, bookmarks, attachments, Factur-X, flatten, optimize, watermark, stamp, rotate and encrypt, and operational routes for health, version, debug, webhooks and remote downloads. The scope names the four conversion routes with their exact paths and uses the page own such as wording for office documents, so no universal format coverage or fidelity claim is made. Screenshot routes are excluded because they return images, and the PDF engine manipulation routes are excluded because they belong to manipulation and would double count the settled edit scope or other jobs. No OCR, editing, compression or cloud processing is claimed, and the surface is the self hosted HTTP API described rather than a hosted service. The settled edit scope, the MIT licence, the open source and locality attributes, pricing and platform findings are unchanged.

  190. 7a053ac

    The Gotenberg record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as offering an ordering control, as preserving bookmarks or metadata, as lossless, as including encryption, PDF/A or PDF/UA, or as covering the optional PDF engine features documented on the same route.
    What it says now
    The job keeps merge and gains the scope merging multiple PDF files into a single PDF with Gotenberg at the /forms/pdfengines/merge route, where the official page states that files are combined in alphanumeric order, numbers first then alphabetical, and that since version 8.37.0 files sharing a filename are all kept and merge in upload order, the route taking a required list of PDF files and returning the merged PDF, with an optional output filename header that defaults to a randomly generated UUID name, from https://gotenberg.dev/docs/manipulate-pdfs/merge-pdfs read on 2026-10-16. The page states the ordering rule explicitly, numbers first then alphabetical, and records that before 8.37.0 only one of several files sharing a filename reached the output while since then all are kept in upload order. Its Basics section gives the POST route, the Gotenberg-Output-Filename header which defaults to a random UUID with the extension appended by Gotenberg, the Gotenberg-Trace header, the required files file list, and a 200 response that is the merged PDF. The same page documents optional features on the route, metadata, bookmarks with autoIndexBookmarks and titleBookmarks and a bookmarks map, attachments with embedsMetadata, Factur-X, flatten, watermark, stamp, rotate and optimize images, and all of those are excluded as separate capabilities: encryption belongs to the unscoped protect job, PDF/A and PDF/UA are manipulation rather than merge, and the remainder are their own operations. Optimize images is not treated as compression and no editing or compression claim is made. No fidelity or losslessness claim is made, since the page describes combining files without promising preservation. The surface is the self hosted HTTP API, not a hosted service. The settled convert and edit scopes, the MIT licence, the open source and locality attributes, pricing and platform findings are unchanged.

  191. 7a053ac

    The Gotenberg record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as promising a particular cipher, as producing permission restrictions that are cryptographically enforced, as offering owner only encryption on every engine, or as including signing, redaction, DRM or legal confidentiality.
    What it says now
    The job keeps protect and gains the scope encrypting PDF files with Gotenberg at the /forms/pdfengines/encrypt route, where the page documents a user password required to open the PDF and an owner password that grants full access and lifts the permission restrictions and defaults to the user password, an owner password alone producing an owner-only PDF since version 8.34.0, allow flags for printing, copying, modifying, annotating, filling forms and assembling pages, and a response that is the encrypted PDF or a ZIP archive when several input files are given, with the page stating that permission restrictions are advisory and not cryptographically enforced once the document opens, and that encryption strength such as AES-256 and permission handling depend on the active PDF engine, with QPDF honouring each permission individually, pdfcpu restricting all permissions if any one is denied, and PDFtk supporting neither owner-only encryption nor permission restrictions, from https://gotenberg.dev/docs/manipulate-pdfs/encrypt-pdfs read on 2026-10-16. The page explains the user and owner password roles, the owner only behaviour since 8.34.0, the requirement that one of the two passwords be set for restrictions or Gotenberg returns 400, and states outright that restrictions are advisory, honoured by viewers but not cryptographically enforced once the document opens. It also states that encryption strength such as AES-256 and permission handling depend on the active PDF engine, with the different behaviour of QPDF, pdfcpu and PDFtk spelled out, and the Basics section documents the route, headers, password fields, the six allow flags defaulting to true, the required files list, and a response that is the encrypted PDF or a ZIP archive for multiple inputs. The scope makes no algorithm guarantee because AES-256 is given only as an example and strength is engine dependent, and it makes no signing, redaction, DRM, key management or legal confidentiality claim. The advisory wording is carried so a reader is not told the restrictions are enforceable. PDF/A, editing and merge are separate routes or jobs and are not folded in, and the surface is the self hosted HTTP API. The settled convert, merge and edit scopes, the MIT licence, the open source and locality attributes, pricing and platform findings are unchanged.

  192. 7a053ac

    The Google Cloud Vision and Document AI record, in the primary ocr job

    What it said
    The ocr job carried no scope, so it could be read as including handwriting, as an accuracy claim, as searchable PDF or PDF/A output, as language breadth, as covering layout parsing or other Document AI processors, or as a single OCR path rather than the two products the record actually combines.
    What it says now
    The job keeps ocr, remains primary, and gains the scope detecting and extracting text with Google Cloud OCR, where the documentation separates Cloud Vision text detection on images and on PDF and TIFF files, including online synchronous small batch annotation of PDF, TIFF and GIF files and the TEXT_DETECTION and DOCUMENT_TEXT_DETECTION features, from Document AI Enterprise Document OCR, which detects and extracts text and layout information from PDFs, scanned document images and Microsoft DocX files with configurable image quality score, language hints and rotation correction, and optional add-ons for math formulas, font style information and checkbox extraction that require Enterprise Document OCR version 2.0 or later, with Cloud Vision text detection described as a globally available REST API with a default quota of 1,800 requests per minute, from https://docs.cloud.google.com/vision/docs/ocr, https://docs.cloud.google.com/vision/docs/pdf and https://docs.cloud.google.com/document-ai/docs/enterprise-document-ocr read on 2026-10-16. The Vision OCR page describes text detection from images and dense document text detection, and the Vision navigation separates image OCR, handwriting OCR and PDF and TIFF OCR. The PDF and TIFF page states the online synchronous small batch annotation support for PDF, TIFF and GIF and names the TEXT_DETECTION and DOCUMENT_TEXT_DETECTION features. The Enterprise Document OCR page says it detects and extracts text and layout information, describes its advanced features as image quality score, language hints and rotation correction, lists PDFs, scanned document images and Microsoft DocX files as the inputs, and describes the OCR add-ons for math formulas, font style information and checkbox extraction as compatible only with version 2.0 and later. Its comparison table also gives Cloud Vision text detection a default quota of 1,800 requests per minute. The scope leaves handwriting out because it is a separate documented page, makes no accuracy claim and treats the image quality score as a documented output rather than an accuracy metric, claims no searchable PDF or PDF/A, no redaction and no editing, and does not duplicate the settled extract job or fold in the Document AI layout parser or other processors. Language hints are named without claiming language breadth since no language list was read, and cloud processing is not inferred beyond the documented service, matching the existing processesLocally attribute. The settled extract scope, attributes, pricing and platform findings are unchanged.

  193. 7a053ac

    The Google Chrome built-in PDF viewer record, in the primary view job

    What it said
    The view job carried no scope, so it could be read as including the annotation, text, form and signing actions listed on the same help page, as including the on device OCR the same page documents, as covering printing or saving to Drive, or as making a mobile claim the Computer tab does not support.
    What it says now
    The job keeps view, remains primary, and gains the scope viewing PDFs in the Chrome PDF viewer on a computer, where the Chrome Help article documents opening a PDF in Chrome and opening it in a new tab, managing the page display with Zoom in and Zoom out, Fit to page and Fit to width, and Rotate for the page orientation, navigating with a setting that turns on navigating pages with a text cursor, caret browsing toggled with F7, tab focus on the PDF, and arrow key navigation by character or word with Shift to select, and a setting at chrome://settings/content/pdfDocuments that chooses whether PDFs open in Chrome or download to the device, from https://support.google.com/chrome/answer/16215622?hl=en&co=GENIE.Platform%3DDesktop read on 2026-10-16 through the repository renderer, because Google Help truncates its body for a plain fetch. The Computer tab documents opening a PDF and opening it in a new tab, and the Navigate a PDF section documents zoom in and zoom out, fit to page and fit to width, rotate, the text cursor page navigation setting, caret browsing with F7, tab focus, arrow key and word navigation with the Windows Linux ChromeOS versus macOS difference, and Shift selection, plus a tip naming chrome://settings/content/pdfDocuments as the open or download setting. The same page also lists highlighting, annotating, drawing, adding text, filling out and signing paperwork and taking notes, and states that scanned PDFs are automatically made searchable and selectable using OCR on the device. Those are excluded from the view scope: the annotation, text, form and signing actions belong to the settled edit and sign jobs, and OCR belongs to the separate unscoped ocr job. Saving to Drive and printing are excluded, no shortcut beyond those stated is claimed, and the platform boundary is a computer because only the Computer tab was read, so no Android or mobile claim is made despite the record platform list. No feature is inferred from the Chrome or PDFium names. The settled edit and sign scopes, attributes, pricing and platform findings are unchanged.

  194. 7a053ac

    The Google Chrome built-in PDF viewer record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as producing a searchable PDF file, as having recognition settings or language options, as an accuracy or handwriting claim, as including the annotation and signing actions listed nearby, or as a mobile capability.
    What it says now
    The job keeps ocr and gains the scope converting scanned PDFs of physical documents to searchable and selectable text in the Chrome PDF viewer, where the Chrome Help article states that scanned PDFs accessed through the viewer are automatically converted using Optical Character Recognition, that after the conversion the text can be highlighted, searched, copied, pasted and found within the document, and that the conversion is completed on the device without sending any data to Google or third parties, from https://support.google.com/chrome/answer/16215622?hl=en&co=GENIE.Platform%3DDesktop read on 2026-10-16 through the repository renderer, re-rendered in this pass to quote the paragraph. A page-wide search of the rendered article returned no occurrence of language, handwriting, accuracy or PDF/A, which is why none of those is claimed. The article describes the conversion as automatic, so the scope claims no recognition settings and no control beyond searching, copying, pasting, highlighting and finding the resulting text, and it does not claim a searchable PDF file as output. The viewer display and keyboard navigation behaviour settled in the previous pass is not duplicated here, and the highlight, annotate, draw, add text, fill out and sign actions listed on the same page stay with the settled edit and sign jobs. Saving to Drive is excluded as cloud storage. The article has both a Computer tab and an Android phone and tablet tab, and only the Computer tab was read, so no Android specific claim is made. The settled view, edit and sign scopes, attributes, pricing and platform findings are unchanged.

  195. 7a053ac

    The GROBID record, in the primary extract job

    What it said
    The extract job carried no scope, so it could be read as including OCR of scanned PDFs, as a conversion or PDF generation tool, as promising layout or bibliographic accuracy, as producing Markdown or JSON from the server, as covering the training and patent services, or as a hosted service.
    What it says now
    The job keeps extract, remains primary, and gains the scope extracting and parsing structured data from PDF documents with GROBID, where the documentation describes it as a machine learning library that extracts, parses and restructures raw documents such as PDF into structured XML TEI encoded documents with a focus on technical and scientific publications, naming header extraction for title, abstract, authors, affiliations and keywords, references extraction and parsing, citation context recognition, full text extraction and structuring, PDF coordinates for bounding boxes and parsing of names, affiliations, addresses and dates, served through the REST endpoints /api/processHeaderDocument, /api/processFulltextDocument and /api/processReferences, which each take a PDF as a required multipart input and return TEI XML or BibTeX through the Accept header, with optional consolidation against Crossref or biblio-glutton and a documented 204 response when no content could be extracted, and with the error reference stating that a PDF with no text blocks, such as a scanned document without OCR, cannot be processed, from https://grobid.readthedocs.io/en/latest/Introduction/ and https://grobid.readthedocs.io/en/latest/Grobid-service/ read on 2026-10-16. The Introduction defines GROBID and describes the library and its functionality list, including header, reference, citation context, full text, PDF coordinate, name, affiliation, address, date, consolidation, patent and funder extraction. The REST API page labels the main services PDF to TEI conversion services, documents the three processing endpoints with their required input PDF and TEI or BibTeX responses, lists the optional consolidation, page range, coordinate, sentence segmentation, ID and flavor parameters, and gives the response codes including 204 for completed processing with no content extracted and 503 when threads are exhausted. It states that the server always returns TEI and that the Markdown and JSON converters live in the Python client, run client side and are lossy. Its error table states that NO_BLOCKS means a PDF with no text blocks, likely a scanned document without OCR, and advises ensuring selectable text and considering OCR, which is why OCR is excluded here. The scope claims no conversion, PDF generation or generic XML conversion beyond TEI and BibTeX, no layout accuracy or bibliographic correctness guarantee despite the reported F1 scores and benchmarks, and does not fold in the training web API, PDF annotation services, patent services or the deprecated batch and Java library pages. No cloud processing is claimed because the service is self hosted and documented through Docker, Gradle and a standalone install. The Introduction says GROBID should run out of the box on Linux 64 bit and macOS, and the credits mention former Windows support; that is recorded here as an observation and the record platform list was left unchanged because the documentation does not state plainly that Windows is unsupported. The settled attributes, pricing and other fields are unchanged.

  196. 7a053ac

    The GroupDocs record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as a family wide claim across all GroupDocs components, as universal format coverage from the 30 or more supported formats line, as a size guarantee, as a fidelity promise, as including encryption which the protect scope already covers, or as a hosted service.
    What it says now
    The job keeps edit, remains primary, and gains the scope editing PDF documents programmatically with GroupDocs.Editor for .NET, where the official tutorial documents loading a PDF through PdfLoadOptions with an optional password into an Editor instance, generating an editable intermediate document as HTML-like content through PdfEditOptions with pagination enabled to preserve page breaks, replacing text, images and layout elements in that editable content, and saving the result back to PDF through the save options, in a .NET 6 or later environment, with a commercial licence required for non trial deployments, from https://tutorials.groupdocs.com/editor/net/document-processing/work-pdf-documents/ read on 2026-10-16. The tutorial defines programmatic PDF editing as modifying content through code and says the API lets you replace text, images and layout elements from C#, then documents the PdfLoadOptions password and memory cache settings, the Editor construction from a stream and load options, PdfEditOptions with EnablePagination preserving original page breaks, Editor.Edit returning an EditableDocument with HTML content and resources, modifying and rebuilding that content, and saving through document save options, with a .NET 6 or later prerequisite and a commercial licence required for non trial deployments. The scope names GroupDocs.Editor for .NET because the settled scopes on this record name GroupDocs.Merger for protect and merge and a parser component for extract, so it makes no whole family claim. It makes no format coverage claim from the 30 or more formats line and no size claim, because the page states both over 500 MB and up to 500 MB, and that discrepancy is recorded here rather than resolved by guess. No fidelity guarantee is claimed beyond the documented pagination behaviour, and the intermediate HTML form is described as editable content rather than a conversion feature. Encryption in the save options is left to the settled protect scope, and OCR, signing, redaction, merging and generic authoring are not claimed. No cloud processing is claimed because this is a self hosted .NET library. The settled protect, extract and merge scopes, attributes, pricing and platform findings are unchanged.

  197. 7a053ac

    The GroupDocs record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as the family wide conversion claim of up to 10,000 combinations, as covering OCR, watermarks or page range options, as a fidelity guarantee, as applying to any platform, or as requiring Office to be installed.
    What it says now
    The job keeps convert and gains the scope converting PDF files with GroupDocs.Conversion for .NET, where the documentation states that the component converts PDF documents to other file formats and also converts other formats to PDF, and its PDF conversion page documents conversions from PDF to Word, Excel, PowerPoint, HTML and images including PNG, JPG and TIFF, converting a single page to an image, and PDF to the eBook formats AZW3, EPUB and MOBI, running on Windows, Linux and macOS with .NET Framework 4.6.2 or .NET 6.0 and later, and requiring no installation of Microsoft Word, Excel or PowerPoint, from https://docs.groupdocs.com/conversion/net/convert/pdf/ and https://docs.groupdocs.com/conversion/net/system-requirements/ read on 2026-10-16. The PDF page states both directions, PDF to other formats and other formats to PDF, and its contents and tables name Word, Excel, PowerPoint, HTML, image formats including PNG, JPG and TIFF, single page to image conversion, and the eBook formats AZW3, EPUB and MOBI. The system requirements page lists the supported Windows, Linux and macOS versions, the .NET Framework 4.6.2 and .NET 6.0 framework floor, and states that no external software such as Microsoft Word, Excel or PowerPoint is required. The docs home page figure of up to 10,000 conversion combinations is deliberately not carried, so no universal format coverage is claimed, and only the PDF directions named on the PDF conversion page are listed. No fidelity or accuracy guarantee is claimed. The image with OCR page and the advanced options pages for watermarks and page ranges are separate capabilities and were not read, so they are not folded in. Editing, merging, signing, viewing and cloud processing are separate components or products and are not claimed, and the settled edit, protect, extract and merge scopes are untouched. The licence position was not re-read and remains the record existing attribute for a commercial per component or GroupDocs.Total licence with free trials.

  198. 7a053ac

    The GroupDocs record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as covering ordinary visual or electronic signature types, as claiming XAdES or PAdES on PDF, as a timestamping or identity assurance service, as a legal validity or compliance claim, as existing signature verification, or as the hosted online apps.
    What it says now
    The job keeps sign and gains the scope signing PDF documents with GroupDocs.Signature for .NET using a digital signature based on an existing PFX certificate, where the documentation states that digital signatures are created from existing PFX certificates containing a private key for signing and a public key for verification, that DigitalSignOptions takes the certificate from a file path or a stream together with its password, plus optional contact, reason and location descriptions and a visible flag deciding whether the signature appears on the page, and that the Sign method writes the signed PDF, with the component running on Windows, Linux and macOS on .NET Framework 4.6.2 or .NET 6, 8 or 10 and requiring no external software, and with the XAdES type property documented as applying to spreadsheet documents only and having no effect on PDF, from https://docs.groupdocs.com/signature/net/sign-document-with-digital-signature.md, https://docs.groupdocs.com/signature/net/sign-document-with-xades-digital-signature.md and https://docs.groupdocs.com/signature/net/system-requirements.md read on 2026-10-16. The digital signature page defines the signature as certificate based with private keys for signing and public keys for verification, says the library creates digital signatures from existing PFX certificates, documents the DigitalSignOptions certificate file path or stream, password, contact, reason, location and visible properties, and gives a PDF example signing sample.pdf with certificate.pfx and writing sampleSigned.pdf. It also states that the XAdES type applies to spreadsheet documents only and has no effect on PDF, and the XAdES page repeats that the feature is only for spreadsheet document types, so XAdES is excluded from PDF by the vendor and the exclusion is stated in the scope. The system requirements page gives Windows, Linux and macOS, .NET Framework 4.6.2 and .NET 6, 8 and 10, separate builds per framework, no external software, and the note that the .NET Framework build runs on Windows only. The page security assertions about authenticity, integrity and non repudiation are not carried as identity or legal guarantees, and no identity assurance, legal validity, qualified signature, eIDAS, compliance, timestamp or PAdES claim is made because none was documented. The text, image, barcode, QR, stamp and metadata signature types and the search and verification features were not read in this pass and are not folded in. The free online apps are a separate hosted offering and are not claimed. Editing, conversion, merging, viewing, OCR, redaction and cloud processing are separate components or products, and the settled edit, convert, protect, extract and merge scopes are untouched.

  199. 7a053ac

    The GroupDocs record, in the view job

    What it said
    The view job carried no scope, so it could be read as rendering to PDF or PDF/A, as printing support, as text extraction, as a browser or hosted viewer product, as a fidelity guarantee, or as covering formats the PDF page does not list.
    What it says now
    The job keeps view and gains the scope rendering PDF documents with GroupDocs.Viewer for .NET, where the official Render PDF documents page states that the component renders PDF files to HTML, PNG and JPEG and that it can be used to implement a simple PDF viewer in a .NET web or desktop application, documents that a Viewer instance takes a document from a file or stream and that the View method renders the whole document or specific pages, and describes HTML output with embedded or external resources and a fixed layout that is the default for PDF or a fluid layout in which graphic content such as images is ignored, PNG and JPG output with selectable width and height or the original page size, rendering of text comments and popup notes, and a TryRepair load option added in version 24.5 that is off by default for structurally corrupted PDFs, with the supported page layout input formats listed as PDF, XPS, OXPS, TEX and OFD with automatic format detection from the file header, running on Windows, Linux and macOS on .NET Framework 4.6.2 or .NET 6 and later without external software such as Microsoft Office, from https://docs.groupdocs.com/viewer/net/render-pdf-documents.md and https://docs.groupdocs.com/viewer/net/system-requirements.md read on 2026-10-16. The PDF page documents HTML, PNG and JPEG output with the HtmlViewOptions, PngViewOptions and JpgViewOptions classes, the embedded and external resource HTML modes, the fixed layout default and the fluid layout caveat that images are ignored, RenderComments for text comments and RenderNotes for popup annotations since 26.8, and the LoadOptions.TryRepair flag that is off by default since 24.5. It lists PDF, XPS, OXPS, TEX and OFD as the supported page layout input formats with automatic detection. The system requirements page gives the Windows, Linux and macOS support, the .NET Framework 4.6.2 and .NET 6 floor, the TFM specific packages, and the statement that no external software such as Microsoft Office is required. The scope makes no PDF or PDF/A output claim, because the PDF input page documents HTML, PNG and JPEG output only, even though the documentation home page lists PDF among rendering formats generally. It claims no printing feature despite the PrintingAllowed value returned by GetViewInfo, no text extraction because ExtractText belongs to the settled extract scope through GroupDocs.Parser, and no browser, hosted or cloud viewer, since the online app links are a separate offering. No fidelity guarantee is claimed and the fluid layout caveat is carried because it changes what a reader would expect. Conversion, editing, merging, signing, OCR and cloud processing are separate components and are not claimed. The system requirements note that a SimSun font must be installed to render documents containing Chinese characters without replacement characters is a rendering dependency recorded here rather than a scope claim. The settled edit, convert, protect, sign, extract and merge scopes, attributes, pricing and platform findings are unchanged.

  200. 7a053ac

    The ABBYY FineReader PDF record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a certificate only digital signature tool, as covering identity assurance or legal validity, as naming a signature standard, as signing on every platform the record lists, or as including cloud signing.
    What it says now
    The job keeps sign and gains the scope signing PDF documents in the PDF Editor of ABBYY FineReader PDF, where the FineReader 16 User Guide documents three separate kinds of signature, a digital signature placed on the page and added from a certificate chosen in a Sign as drop down list, with certificate properties viewing, a certificate installation path through the Windows Certificate Import Wizard, optional reason, location and contact fields, selectable signature details, an optional time stamp server URL with login and password, borders and background and an optional custom image, saved with Save and sign, where adding a signature to an already signed document keeps the previous signatures valid and more than one digital signature can be added, a text signature created from text with a chosen style and then placed, resized, rotated or moved on the page, and a picture signature created from an image file or the clipboard and placed the same way, and where the guide states that opening a PDF automatically validates the digital signatures in it and shows the results in a Digital Signatures pane, that signature properties can be viewed, and that a signature can only be removed by someone holding the corresponding digital certificate, from https://help.abbyy.com/en-us/finereader/16/user_guide/digitalsignature, https://help.abbyy.com/en-us/finereader/16/user_guide/textsignature and https://help.abbyy.com/en-us/finereader/16/user_guide/picturesignature read on 2026-10-16 through the repository renderer, because the pages sit behind a Terms of Use overlay and return only markup to a plain fetch. The Digital signature page documents the Add Digital Signature flow, the certificate Sign as list, Show Certificate, installing a certificate through Windows Explorer and the Certificate Import Wizard, the Reason, Location and Contact fields, the selectable displayed details, the optional time stamp server URL with credentials, borders and background, the optional custom image, Save and sign, the note that a signature records the document version at signing and that adding a signature keeps previous signatures valid, Show Signature Properties, automatic validation on opening a document with results in the Digital Signatures pane, removal only with the corresponding certificate, and support for more than one signature. The Text signature page documents typed text signatures with a style, placed and resized, rotated or moved, and the Picture signature page documents image or clipboard signatures with size changes and a transparent background, placed the same way. No identity assurance, legal validity, qualified signature, eIDAS, PAdES, XAdES, timestamp standard, certificate authority, algorithm or key length claim is made. The vendor statements that a digital signature is used to establish identity and to show the document was not altered are recorded here as vendor wording rather than as guarantees. The documented steps rely on Windows Explorer and the Windows Certificate Import Wizard, and the scope names them as the guide does without asserting a platform policy, so no macOS or iOS signing claim is made. Editing, protection, OCR, conversion, merging, viewing and cloud storage are separate jobs or features and are not folded in. The settled ocr, convert, edit and protect scopes, attributes, pricing and platform findings are unchanged.

  201. 7a053ac

    The Apple Preview record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as a bulk file merge tool, as lossless, as silent about the auto save behaviour, as applying on iOS, or as including password protected documents and cloud actions.
    What it says now
    The job keeps merge and gains the scope combining PDF pages and documents in the Preview app on Mac, where the Apple Preview User Guide states that you can add individual pages from one PDF to another and merge the entire contents of two or more PDFs by opening the PDFs, choosing View Thumbnails and dragging thumbnails from the sidebar of one PDF to the sidebar of the other, adding pages at the end of a document or between pages and dragging thumbnails to reorder them, or by dragging a PDF icon from the Finder onto the sidebar of an open document to add a whole PDF, and where its companion page documents inserting a page from a file or a blank page after the current page, moving pages by dragging, deleting selected pages with the Delete key, and moving a page from one PDF to another by dragging thumbnails, with the notes that deleting a page also removes all annotations on it, that a new PDF can be created by dragging thumbnails from a PDF sidebar to the desktop, and that changes to a Preview document are saved automatically, so the original PDFs should be duplicated first with File Duplicate to keep copies, from https://support.apple.com/en-us/guide/preview/prvw43696/mac and https://support.apple.com/en-us/guide/preview/prvw11793/mac read on 2026-10-16 through the repository renderer, because Apple guide pages return an unsupported content type to a plain fetch. The combine page states both the single page and whole document cases, the View Thumbnails and drag workflow, the Command click selection, adding pages at the end or between pages, dragging to reorder, dragging a PDF icon from the Finder to the sidebar, the auto save warning and the File Duplicate advice. The add, delete or move page states the Edit then Insert Page from File or Blank Page insertion after the current page, the sidebar insert control tip, deletion with Delete or Edit then Delete together with the note that annotations on the deleted page are removed, page reordering by dragging, moving pages between PDFs by dragging thumbnails, and creating a new PDF by dragging thumbnails to the desktop. The scope is limited to the Preview app on Mac as the guide titles and describes it, so no iOS claim is made despite the record platform list, and no claim is made about merging without an open document beyond the documented Finder drag or about password protected documents, which the pages do not mention. No fidelity, quality or lossless claim is made, and the auto save behaviour and annotation loss on deletion are carried because they change expectations. The annotation and markup capabilities belong to the settled edit job, viewing, signing, password protection, conversion, compression and printing are settled jobs or separate features, and no iCloud or other cloud action is claimed. No operation is inferred from thumbnails or drag and drop: every one named is stated by the guide. The settled view, sign, edit, protect, convert and compress scopes, attributes, pricing and platform findings are unchanged.

  202. 7a053ac

    The CamScanner record, in the compress job, refreshed retrieval result

    What it said
    The earlier blocker recorded only two first party surfaces and said a dedicated compression or toolbox help page would be needed to settle the job. This entry records the further surfaces checked so the blocker rests on an exhaustive sweep rather than on two pages, and so that no later pass repeats the same retrieval.
    What it says now
    Still unresolved, and now checked against four first party surfaces on 2026-10-16, each read directly or through the repository own headless renderer with renderer exit 0. First, https://www.camscanner.com/compress-pdf returns an Error 404 page of 238 characters, so no such toolbox route exists. Second, the official homepage at https://www.camscanner.com/ was rendered and its toolbox navigation enumerated from the rendered document: the tools are PDF to Word, Word to PDF, PDF to Excel, Excel to PDF, PDF to PPT, PPT to PDF, PDF to JPG, JPG to PDF, Merge PDF, Split PDF pointing at extract-pdf-pages, Organize PDF pointing at sort-pdf-pages, Rotate PDF Pages, Delete PDF Pages, Add Watermark, Protect PDF and Sign PDF, seventeen entries in all, and none of them compresses a PDF. In the homepage visible text the word compress appears exactly once and only inside a customer testimonial, a teacher quoted as saying adding annotations, and merging, splitting, and compressing files are all a breeze, which is not vendor documentation of an operation, an input, an output or a limit. Third, the official agent documentation capability index at https://www.camscanner.com/agent-docs/capabilities/ was rendered and lists eight categories, scan and image enhancement, text recognition and structured extraction, document format conversion, multi-page and batch processing, translation and content editing, watermarks and authenticity detection, and cloud saving and document management, with no compression category and no occurrence of the word compress, and the conversion capability page at https://www.camscanner.com/agent-docs/capabilities/conversion/ likewise contains no occurrence of compress, file size or reduce. Fourth, the official CamScanner Skill definition at https://raw.githubusercontent.com/CS-CamScanner/camscanner/main/SKILL.md, version 1.1.8, states its purpose as document conversion and processing and gives a complete command table of twenty eight commands across image enhancement, format conversion, watermark, translation, formula, merge, PDF, recognition, detection, editing, receipt and cloud document groups, and that table contains no compression command of any kind. Its only compression wording is in the error handling table, where the remedy for a file size exceeds the maximum limit error is the instruction to compress the file and retry, which tells the reader to compress externally rather than offering a CamScanner compression operation. The skill also documents image hd as upscaling and image enhance as shadow removal, sharpening and black and white conversion, which are image processing and not PDF size reduction, so they are not evidence for this job. The job is therefore left unresolved rather than scoped from scan resolution, from the word optimize, or from a testimonial. It is not removed because nothing contradicts it: no first party page says compression is unavailable, the sources simply never document it. Settling it would need a first party page or CLI command that names a compression operation on an existing PDF together with its input, output and any size or quality control.

  203. 7a053ac

    The camscanner compress ledger commit, in its page-lastmod change

    What it said
    The commit message said the regenerated page-lastmod file was included so the corrections page hash would match its content, which implies a single page entry changed. In fact that commit rewrote the stored hash for all 209 pages while leaving every date unchanged.
    What it says now
    Recorded accurately here. The commit also rewrote all 209 stored hashes with no date changes. The question was whether that was a real content change or a stale derived file, and the checks done afterwards point to the second: a rebuild of the current tree reproduces the committed map exactly and the build reports 209 URLs seen, 209 dated, 0 dates moved; the only per-build value in a page is the dialegein-build meta element, whose whole content attribute is replaced with a fixed token before hashing, and the commit id appears nowhere else in a built page, so the hash cannot depend on the commit; and recomputing the normalised SHA-256 of the built /about/ page by hand reproduces the value now stored for /about/ exactly. The 209 hash rewrite therefore means the map committed before that commit did not match a fresh build of the current tree, which is stale derived data rather than a content change, and no page date moved as a result. The commit message cannot be corrected in place without rewriting history, so the accurate account is recorded here instead. One tooling observation follows and is not a request for a change: verify does not catch this class of drift, because check-lastmod validates that every URL carries a date the map can account for and does not recompute hashes, so a build and a diff would be needed to notice a stale map. The map as committed is reproducible.

  204. 7a053ac

    The CamScanner record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as editing existing PDF documents, as treating annotation markup as content editing, or as covering page organisation, crop, enhancement or export. The earlier blocker in c62f582 recorded that no first party surface documented editing existing PDF content and left the job unscoped.
    What it says now
    The job keeps edit and gains the scope editing text and areas in images with CamScanner, where the official CamScanner skill and its Image Processing Reference describe a three step image scan then image edit flow that accurately replaces, deletes or moves text in an image while preserving the original layout and visual style, working from the character level OCR indexes returned by the scan step, with an update request replacing a character index range with new text, a delete request removing a text, table, image or stamp area, and a move request placing an area at a target position, producing an edited image file, and where the documented inputs are images in JPG, JPEG or PNG only and PDF inputs are routed to the PDF commands, so editing an existing PDF document is not documented, from https://raw.githubusercontent.com/CS-CamScanner/camscanner/main/SKILL.md and https://raw.githubusercontent.com/CS-CamScanner/camscanner/main/references/image-processing.md read on 2026-10-16. The skill command table gives image scan as layout analysis returning character indexes and image edit as replace, delete or move text based on scan results with an image output, and the reference page documents the scan then edit flow, the update, delete and move edit requests with their parameters and area types, the character index example replacing East with West, and writing the result to an edited jpg. SKILL.md lists JPG, JPEG and PNG as the supported image formats and routes PDF inputs to the pdf commands, and its command table has no PDF editing command. This supersedes the earlier blocker in c62f582, which had examined the Play listing, the 404 edit-pdf route and the homepage navigation but not the official skill reference. The scope claims image text and area editing only and states that editing an existing PDF document is not documented, so it does not read the annotation tools as content editing and does not claim page organisation, crop, enhancement, scan editing or export, which are separate features or jobs. The layout and style preservation wording is attributed as the reference own description rather than an accuracy guarantee, and no handwriting, OCR accuracy, identity or cloud claim is added beyond the documented CLI cloud save behaviour. The settled ocr, convert and protect scopes and the compress and sign blockers are otherwise untouched.

  205. 7a053ac

    The CamScanner record, in the sign job

    What it said
    The sign job carried no scope. The earlier blocker in c62f582 read the same Sign PDF page, recorded that it documents a drawn or uploaded signature image rather than a cryptographic signature, and left the job unresolved on the reasoning that the catalog holds signing to a cryptographic standard, while noting that a decision that a drawn signature satisfies the job would settle it.
    What it says now
    The job keeps sign and gains the scope adding a signature to PDF documents with the CamScanner web Sign PDF tool, where the official page says you add your signature to PDF documents online and then save the signed file for sharing or recordkeeping, and describes the mark as a signature drawn by hand or uploaded as an image, with seals, logos or cross page seals added when a document needs more than a signature, signing a document yourself or sending it to other people for their signatures with a due date, and importing a file or opening a document from the CamScanner cloud to begin signing, with processing stated to happen on cloud servers, and where the page contains no wording about certificates, digital signatures, timestamps, identity verification or an audit trail, so the documented operation is a visual signature rather than a certificate based digital signature, from https://www.camscanner.com/sign-pdf read on 2026-10-16 through the repository renderer. The decision to scope rather than keep blocking rests on the settled scopes of other records: Apple Preview sign is scoped as a drawn signature, and ABBYY FineReader PDF sign is scoped to three signature kinds with the mechanism named for each, so a drawn signature with its mechanism stated satisfies the job. The page documents drawing a signature by hand or uploading an image, adding seals, logos or cross page seals, signing for others with a due date, importing a file or opening a cloud document to begin signing, and cloud based processing on servers that add signatures to PDF files. A case sensitive search of the rendered text found no occurrence of certificate, digital signature, PAdES, eIDAS, timestamp, identity or audit, so no certificate, digital signature, qualified signature, timestamp, identity assurance, legal validity, compliance or audit trail claim is made. The 256-bit SSL wording is transport security and is not turned into a document protection claim, which belongs to the settled protect scope. The invite and due date wording is recorded as stated and not described as a certificate based signature request. Image editing, annotations, protection, merge, OCR and conversion are separate settled jobs or blockers and are not folded in. The settled ocr, convert, edit and protect scopes and the compress blocker are unchanged.

  206. 7a053ac

    The CommonLook PDF record, in the edit job, refreshed retrieval result

    What it said
    The earlier exclusion rested on the Allyant product page and reasoned from the absence of content editing vocabulary there. This entry records the official product manual, which states the boundary outright, so the exclusion rests on a direct vendor statement rather than on absent wording.
    What it says now
    Still unresolved, and the exclusion is now stated by the vendor. The official CommonLook PDF User Guide at https://manuals.commonlook.com/resources/manuals/commonlook-pdf-global-access-user-guide/, reached from the CommonLook Manuals index at https://manuals.commonlook.com/ after the old commonlook.com guide paths were found to redirect cross origin to allyant.com and return its page not found response, opens with a Remediator Tip that reads: Using CommonLook PDF to fix your PDFs will not change the physical appearance of the document. That sentence settles the job on its own: editing an existing PDF content means changing what the document shows, and the vendor states that this tool does not. The guide was then checked for any operation that would contradict it. Its Contents are Welcome and Getting Started, The Toolbar and Ribbon, Insert Tag and Convert Tag Tabs, The Settings Tab, The Windows Tab, The Operations Tab, The Tag Properties Panel and The Integrity Validator, which is a structure and validation surface rather than a content editing surface. The Operations Tab page was read in full and lists operations that act on the tag tree: Create Document Tag, Untag, tag properties, expand and collapse, paste, create tag, cleanup, highlight content, flatten tags, artifact all images, remove role maps, merge tags, copy, cut, level up, change tag order, tag similar, copy contents to actual text or alt text, split text, mark font as all caps, calculate bounding box, linearize table, insert row, insert caption, create linked headers, show table editor, merge cells, split to rows, and the insert operations for lists and tables of contents such as insert LI, LBody, Lbl, TOCI, Reference, NonStruct, Paragraph and TOC. Every one of those inserts or rearranges tag structure, and the apparently content bearing ones are tag only: Insert row places a new TR tag, Insert LI places a new empty LI tag, Split Text splits a text run inside the tag tree, and Mark Font as All Caps gives the text alternative text in capitals rather than changing the document appearance, which the Remediator Tip already rules out. The Toolbar page documents Quick Save, Save and Close which compresses an inflated file, Undo and Redo, so its only file level effects are saving and size compression, and compression is a separate job on this record. The documented operations are therefore accessibility remediation and validation, which the accessibility job on this record already carries, and no operation on the pages read changes existing text, images or layout. The job stays unresolved rather than scoped, because scoping it would duplicate the accessibility scope and would tell a reader the tool edits document content, which the vendor denies. It is not removed because the vendor does not say content editing is unavailable as a product category; it says fixing PDFs will not change physical appearance, which is a boundary on the remediation workflow rather than an unsupported operation list. A manual page describing an operation that changes the rendered document would settle it.

  207. 7a053ac

    The ICEpdf record, in the primary view job

    What it said
    The view job carried no scope, so it could be read as including printing, conversion, extraction, annotation editing, signing or OCR, as a current release claim, or as a general claim about modern Java and platform versions rather than the ones the documentation states.
    What it says now
    The job keeps view, remains primary, and gains the scope viewing and rendering PDF documents with ICEpdf, where the official overview describes an open source 100 percent Java PDF library for displaying PDF documents within a Java application that can act as a stand alone PDF viewer or be embedded as a viewer component, and the 4.2 release it documents renders PDFs up to version 1.6 (Acrobat 7.0), with the viewer providing zoom in and out, rotate, next and previous page, fit to window, width or actual size and pan, multipage continuous and side by side view types, a utility pane with outlines and document search, document information and permissions dialogs, text selection and clipboard functions, and search result highlighting on the pages where results occur, documented for Windows, Linux, Solaris and Mac OS X running Sun J2SE 1.5, 1.6 or 1.7, from https://www.icepdf.org/product/overview.html and https://www.icepdf.org/product/supported-platforms.html read on 2026-10-16, reached from https://www.icepdf.org/ after https://icepdf.org/ redirected cross origin to the www host. The overview describes the library as ideal for displaying and printing PDF documents, gives the PDF version 1.6 rendering boundary for release 4.2, and lists the viewer features quoted in the scope; the supported platforms page gives the Windows, Linux, Solaris and Mac OS X grid with the Sun J2SE versions listed there. Printing is documented but excluded, as are rendering to PNG, JPEG or GIF and conversion to SVG, which belong to the conversion job, text and metadata extraction which belong to the extract job, link annotation editing which belongs to the edit job, and signing and OCR which are separate jobs. No fidelity, accuracy, browser, PDF/A or cloud claim is made, and the server side deployment wording is not turned into a hosted service claim. Version context: the official site documents release 4.2 and carries a 2009 copyright, and the record tagline statement that the project is maintained as a community project on GitHub could not be verified, because https://github.com/ICEsoft/icepdf returns 404, so that statement was left unchanged rather than edited on a failed URL guess. The attributes, pricing and platform findings are unchanged.

  208. 7a053ac

    The ICEpdf record, in the extract job

    What it said
    The extract job carried no scope, so it could be read as promising output formats or layout fidelity, as including OCR of scanned PDFs, as a conversion or rendering claim, as a hosted service, or as inheriting the viewer text selection and search features instead of using the library extraction path.
    What it says now
    The job keeps extract and gains the scope extracting content from PDF documents with ICEpdf, where the official overview lists PDF content extraction as using ICEpdf to extract PDF document metadata, text and images, and the deployment scenarios page says ICEpdf can be deployed to client workstations for interactive viewing as well as to headless server environments for non visual PDF content conversion and extraction applications, with the product documented as 100 percent pure Java and usable with any J2SE or J2EE JVM version 1.5.0 or greater, from https://www.icepdf.org/product/overview.html and https://www.icepdf.org/product/deployment-scenarios.html read on 2026-10-16. The overview names metadata, text and images as the extracted content, and the deployment page supplies the headless server extraction context and the JVM floor. The PDF compliance page read in this pass is a table of supported PDF specification features and adds no extraction operation or limitation. The scope makes no claim about output file formats, encoding, layout fidelity, reading order, OCR of scanned documents or accuracy, because none is stated. It does not infer extraction from the viewer text selection and search behaviour, which belongs to the settled view scope. The conversion wording on the deployment page is not claimed beyond extraction, because rendering to images or SVG and converting to other formats belong to the convert job. Editing, signing, printing, PDF/A, browser and cloud claims are not made, and the servlet and headless deployment wording is not turned into a hosted service claim since the product is a library deployed by the user. The version context recorded with the view scope applies here too: the official site documents release 4.2 and carries a 2009 copyright, and these capability statements are the vendor own. The settled view scope and its PDF 1.6 rendering boundary, the attributes, pricing and platform findings are unchanged.

  209. 7a053ac

    The ICEpdf record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage from the phrase other file formats, as a fidelity or lossless claim, as covering text or Office conversion, or as a hosted service rather than a library the user deploys.
    What it says now
    The job keeps convert and gains the scope converting PDF documents to other file formats with ICEpdf, where the official overview says the library can render PDFs to an image file in PNG, JPEG or GIF in a server side web application, and that it can be used to convert rendered PDF pages to images, SVG documents and other file formats, and the deployment scenarios page says the servlet renders the requested files server side, converts them to a specified image format and serves the pages back in the form of a standard web page, with the site documenting release 4.2 and describing the product as 100 percent pure Java, from https://www.icepdf.org/product/overview.html, re-read unchanged in this pass, and https://www.icepdf.org/product/deployment-scenarios.html read on 2026-10-16. The overview names PNG, JPEG and GIF image files and SVG documents explicitly, and the deployment page adds the server side conversion of pages to a specified image format. The scope deliberately does not enumerate the vendor phrase other file formats, so no Office, text, HTML or universal format coverage is claimed. No fidelity, quality, resolution, lossless or layout preservation claim is made, and only rendered PDF pages are claimed as the input to conversion. Extraction belongs to the settled extract job, viewing to the settled view job, and editing, signing, printing, PDF/A, browser and cloud claims are not made. The servlet wording is carried as deployment context and not as a hosted service. The version context recorded for the settled view and extract scopes, release 4.2 with a 2009 copyright, applies here and is stated in the scope as the release the site documents. Checks ran before committing: build exit 0, check-job-scopes exit 0, check-coverage exit 0, check-lastmod exit 0, full verify exit 0.

  210. 7a053ac

    The ICEpdf record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as general PDF content editing of text, images or layout, as any kind of annotation editing rather than link annotations, or as a user facing editing feature rather than a developer callback hook.
    What it says now
    The job keeps edit and gains the scope editing PDF documents with ICEpdf, where the official overview lists support for document editing and saving for Link Annotation edits and separately states support for interactive link annotation actions of the uri, launch and resource kinds, and its PDF Link Annotations section says ICEpdf can be configured to support interactive link annotations via a mouse and that an annotation callback gives developers flexibility in which types of link annotation actions they wish to support, so the documented edit capability is adding and saving link annotations and their actions rather than changing existing text, images or page layout, in a 100 percent pure Java library documented at release 4.2, from https://www.icepdf.org/product/overview.html re-read unchanged on 2026-10-16. The page states document editing and saving for Link Annotation edits, supports interactive link annotation actions of the uri, launch and resource kinds, and describes configuring interactive link annotations via a mouse with an annotation callback giving developers flexibility over which link annotation actions to support. Scoping is consistent with the catalog precedent set by the settled Apple Preview edit scope, annotate, not existing text, and the settled Chrome PDF Viewer edit scope, draw, highlight and a hand drawn signature, which accept annotation level editing when the boundary is stated. The scope states that the capability is adding and saving link annotations and their actions and not changing existing text, images or page layout, because no page content editing appears on the page. Only the three named action kinds are carried, and no other annotation type, markup, text annotation, signing, OCR, conversion, printing, PDF/A, fidelity, browser or cloud claim is made. The annotation callback wording is recorded as a developer hook rather than a user facing feature. The version context recorded for the settled view, extract and convert scopes, release 4.2 with a 2009 copyright, applies here and is stated in the scope.

  211. 7a053ac

    The ICEpdf record, in the sign job, retrieval result

    What it said
    The sign job carried no scope and nothing on this record recorded what the official surfaces say or do not say about signing, so the gap could be mistaken for an unchecked candidate rather than a searched one.
    What it says now
    Still unresolved, and the absence is now documented. The official ICEpdf site at https://www.icepdf.org/ was swept on 2026-10-16 across twelve pages, each fetched with a desktop user agent and returned HTTP 200, and the visible text of every one of them was searched case insensitively for signature, signed, signing, sign PDF, digital signature, certificate, PAdES, eIDAS, timestamp, time stamp and the bare word sign. The pages were the home page, product/overview.html, product/deployment-scenarios.html, product/supported-platforms.html, product/pdf-compliance.html, product/font-engine.html, resources/documentation.html, downloads.html, support/support-options.html, sitemap.html, demos.html and contact-us.html. Every one returned no signing vocabulary at all: zero occurrences, including the bare word. The PDF compliance page was checked in detail because a signing capability usually appears there as a signature field or an annotation type, and it does contain rows for markup, text, free text, line, square and circle, polygon and polyline and text markup annotations and rows for button fields, check boxes, radio buttons, text fields and choice fields, but it has no signature annotation row, no signature field row and no occurrence of widget or signature anywhere in the table. The documentation page offers no route to an API reference either: its release notes and developer guide cells are labelled wiki but carry no href, the only external links on the page are the ICEfaces forum and the corporate site, and the paths api, javadoc, wiki, resources/api.html and resources/javadoc.html all return 404 while docs returns 403. So no first party signing operation, signature field, certificate handling or verification behaviour was found in any official surface that could be reached. The job is left unresolved rather than scoped, because none of this is evidence that signing is supported, and it is not removed because absence of documentation is not a vendor statement that signing is unsupported. The settled view, extract, convert and link annotation edit scopes and the version context recorded with them, release 4.2 and a 2009 copyright, are unchanged. A reachable first party developer guide or API reference, or a product page naming a signature operation, would settle it.

  212. 7a053ac

    The iLovePDF record, in the primary merge job

    What it said
    The merge job carried no scope, so it could be read as a claim about file count or size limits, as preserving metadata or fidelity, as a cloud storage feature, as covering the API desktop and mobile apps, or as including other tools the same site offers.
    What it says now
    The job keeps merge, remains primary, and gains the scope merging PDF files with the iLovePDF web Merge PDF tool, where the official user guide section for Merge PDF says that to merge two or more PDFs you select the documents, that you can arrange the files however you like before merging them, that you can also combine several password protected documents into one, and that you upload the selected files, process them and then download the merged file, and where the Merge PDF tool page says you can combine PDFs in the order you want, with the guide pointing to the Merge PDF limits for each account type instead of stating a single numeric limit, from https://www.ilovepdf.com/help/documentation and https://www.ilovepdf.com/merge_pdf read on 2026-10-16 through the repository renderer, and https://www.ilovepdf.com/help/faq read the same day. The guide states the two or more PDFs input, the free arrangement of the files before merging, the ability to combine several password protected documents into one, the upload, process and download flow, and directs the reader to Merge PDF limits for each account type. The tool page states combining PDFs in the order you want and selecting or dropping PDF files. The FAQ was checked for merge specific limits and gives only general service facts, including that processed files are kept for a maximum of two hours so they can be downloaded, and the general upload methods, with no merge specific count or size limit. Ordering is carried because both sources state it and not inferred from the tool name. No numeric limit is stated because none is given, and the scope records that limits are per account type rather than inventing a figure. Password protected inputs are carried as an input property rather than a protection feature, since protection is the separate protect job. The guide cloud storage input sources are deliberately not carried, so no cloud storage or sync claim is made, and no API, desktop or mobile app claim is made because only the web tool page and the guide were read. Compression, editing, conversion, signing, OCR, viewing, redaction and repair are separate jobs and are not folded in. No fidelity, lossless or metadata preservation claim is made.

  213. 7a053ac

    The iLovePDF record, in the compress job

    What it said
    The compress job carried no scope, so it could be read as a lossless or quality preserving reduction, as offering settings beyond the three documented levels, as stating a numeric size limit, as a cloud storage feature, or as the tool page marketing phrase about maximal PDF quality.
    What it says now
    The job keeps compress and gains the scope compressing PDF files with the iLovePDF web Compress PDF tool, where the official user guide section for Compress PDF says that to compress PDF files you select the documents, upload the file or files to optimize, choose between extreme, high or regular compression level and then hit the Compress PDF button, and that the higher compression, the lower file size and quality, with the button downloading your reduced PDF, and where the Compress PDF tool page says it reduces file size, with the guide pointing to the Compress PDF limits for each account type instead of stating a single numeric limit, from https://www.ilovepdf.com/help/documentation and https://www.ilovepdf.com/compress_pdf read on 2026-10-16 through the repository renderer. The guide names the three levels extreme, high and regular, states the trade off that the higher compression the lower file size and quality, and says the button downloads the reduced PDF or saves it back to the cloud, while directing the reader to the limits for each account type. The tool page says it reduces file size while optimizing for maximal PDF quality and invites selecting or dropping PDF files. The three levels are carried exactly and no resolution, DPI, percentage, target size or other quality setting is claimed. The trade off is carried as the vendor states it and not softened into a lossless or quality preserving claim. The tool page phrase optimizing for maximal PDF quality is marketing wording and is deliberately not carried as a quality guarantee; it is recorded here instead. No numeric limit is invented, because the guide defers limits to each account type. The guide cloud storage sources and the save back to the cloud step are not carried, per the requirement to keep compression separate from cloud storage, and the download of the reduced PDF is the only output behaviour claimed. Merge, edit, conversion, signing, OCR, protection, viewing and generic file size statements are separate jobs or claims and are not folded in, and the settled merge scope is unchanged.

  214. 7a053ac

    The iLovePDF record, in the edit job

    What it said
    The edit job carried no scope, so it could be read as editing existing PDF text, images or page layout, as covering the neighbouring page number, watermark or repair tools, as a cloud storage feature, or as carrying a limits claim the Edit PDF section does not state.
    What it says now
    The job keeps edit and gains the scope adding content to PDF pages with the iLovePDF web Edit PDF tool, where the official user guide section for Edit PDF says that to add elements or text to a page you choose from the Text, Image or Shape icons in the Edit PDF toolbar, that you can resize or move any element on the page where it was created, that you can change the colour, transparency, border color, rotation, format or line thickness of your images, shapes or texts using their property bars, and that you then click the Edit PDF button to process the file, and where the Edit PDF tool page describes it as editing a PDF by adding text, shapes, comments and highlights, so the documented capability is adding and formatting new elements and comments rather than changing existing text, images or page layout, from https://www.ilovepdf.com/help/documentation and https://www.ilovepdf.com/edit-pdf read on 2026-10-16 through the repository renderer. The guide documents choosing Text, Image or Shape from the toolbar, resizing and moving an element on the page where it was created, and the property bar controls for colour, transparency, border color, rotation, format and line thickness, then clicking Edit PDF. The tool page describes the PDF Editor as editing by adding text, shapes, comments and highlights. Every documented action applies to elements the user places, so the scope states that the capability is adding and formatting new elements and comments rather than changing existing text, images or page layout, and makes no claim that existing PDF content can be edited. Only the formatting controls the guide names are carried. Comments and highlights come from the tool page and are carried as additions. The neighbouring guide sections for Add page numbers, Add watermark and Repair PDF are separate tools and jobs and are not folded in, nor are OCR, conversion, signing, protection, viewing, merge or compression. The guide cloud storage input and save back to the cloud steps are not carried, consistent with the settled merge and compress scopes. No limits claim is made, because the Edit PDF section states none unlike the sections around it, and no fidelity or lossless claim is made.

  215. 7a053ac

    The page-lastmod churn explanation recorded in e448a37, observed again

    What it said
    That entry concluded that the hash cannot depend on the commit, because the only per-build value in a page is the dialegein-build meta element whose whole content attribute is normalised to a fixed token and because the commit id appears nowhere else in a built page. It also called the churn stale derived data. During the ilovepdf edit ledger commit the same churn recurred: a commit that added only a corrections entry rewrote all 209 stored hashes while changing no page date.
    What it says now
    The observation stands but the explanation is not established, and this entry corrects the overclaim. What is measured: the ledger commit for the ilovepdf edit scope changed 418 hash lines, which is 209 entries, and zero date lines, and the build that produced it reported 209 dates moved. Against that, a rebuild of the current tree changes nothing at all: two consecutive builds produce pages that are byte identical once the dialegein-build meta content is normalised, the build reports zero dates moved, and the stored hash for /about/ equals the SHA-256 of the normalised built page when the script own replacement token is used. A hand recomputation with a different replacement token gives a different hash, which is why an earlier check of this kind is not evidence about commit dependence. So the map is internally consistent and reproducible against the current build, yet hashes changed across commits whose page content did not change. A dependence of the hash on the commit id therefore cannot be ruled out, and the claim that it can be ruled out is withdrawn here. Establishing the cause would need a controlled comparison, building the same content at two different commits, which was not done and which would have meant rewriting local history. Nothing user facing depends on it: zero sitemap dates changed in either churn, because the dates were already the current date, and verify stays green before and after. This is recorded as a tooling question for a dedicated pass rather than patched again in a content pass.

  216. 7a053ac

    The iLovePDF record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage from the site convert tool list, as OCR of scanned PDFs, as a fidelity or layout guarantee, as stating numeric limits, as covering the neighbouring security and repair tools, or as an editable output where the vendor says non-editable.
    What it says now
    The job keeps convert and gains the scope converting documents between PDF and other formats with the iLovePDF web conversion tools, where the official user guide documents the Convert to PDF tools as JPG to PDF with portrait or landscape orientation, with or without margin, and an option to merge all images into a single PDF, WORD to PDF turning a .docx into a non-editable document, POWERPOINT to PDF turning a .pptx into a non-editable document, EXCEL to PDF turning a .xlsx into a PDF, and HTML to PDF from a pasted URL with orientation and margin options, and the Convert from PDF tools as PDF to JPG including extracting all the pictures in a zip file, PDF to WORD producing an editable .docx, PDF to POWERPOINT producing an editable .pptx where each page becomes a slide, PDF to EXCEL producing an Excel document, and PDF to PDF/A for long term archiving and preservation, with each tool deferring its limits to the account type rather than stating a number, from https://www.ilovepdf.com/help/documentation read on 2026-10-16 through the repository renderer. The guide names the ten directions with their options and outputs, distinguishes non-editable outputs for the Convert to PDF directions from editable outputs for PDF to WORD and PDF to POWERPOINT, and ends each section by deferring limits to the account type. Two sentences are internally inconsistent and are not carried: the PDF to POWERPOINT section says the pictures, text and graphics will be converted to a PDF file, contradicting its own next sentence about a .pptx per slide and a .pptx download, and the PDF to EXCEL section writes .xslx rather than .xlsx; the scope follows the surrounding statements and the discrepancy is recorded here instead. No direction, format or universal coverage beyond the ten named is claimed, so nothing is inferred from the site navigation convert list. OCR is excluded because the guide documents it as a separate tool and the site FAQ says scanned PDF conversion requires OCR, so that belongs to the separate OCR job. No fidelity, layout preservation or accuracy claim is made, and the HTML to PDF preview is not turned into one. No numeric limits are stated because each section defers them to the account type. Editing, merging, compression, signing, protection, viewing and repair are separate tools and jobs and are not folded in, and the guide cloud storage inputs and save back to the cloud steps are not carried, consistent with the settled scopes on this record. PDF to PDF/A is carried as the guide states it, for long term archiving and preservation, with no compliance or conformance claim added.

  217. 7a053ac

    The iLovePDF record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a certificate based digital signature or qualified signature service, as offering signature verification, as a legal validity guarantee, as a timestamp service, as a cloud storage feature, or as including the other tools on the site.
    What it says now
    The job keeps sign and gains the scope signing PDF documents with the iLovePDF web Sign PDF tool, where the official user guide documents signing your own document by uploading it and choosing a signature from the sidebar menu, drawn by hand, uploaded as an existing signature or typed as text, optionally adding fields such as date and initials, then clicking Sign PDF to download the signed PDF, and documents collecting e signatures by creating a signature request in which you add signers, include signing fields, adjust security parameters and send, after which all signers receive an email with access to the document and can read and sign it, and once the process is complete you receive a copy of the signed document with the option to download an audit trail, and where the Sign PDF tool page describes signing a document yourself or sending a signature request to others, and the guide contains no wording about certificates, digital signatures, PAdES, eIDAS or timestamps, so the documented signatures are visual and workflow signatures rather than certificate based digital signatures, from https://www.ilovepdf.com/help/documentation and https://www.ilovepdf.com/sign-pdf read on 2026-10-16 through the repository renderer. The guide documents the three signature input kinds, the optional date and initials fields, the download of the signed PDF, the signature request workflow with signers, signing fields, security parameters, an email giving access to the document, and a completed copy with the option to download an audit trail. A case insensitive search of the guide found no certificate, digital signature, PAdES, eIDAS or timestamp wording, so no cryptographic mechanism is claimed. The vendor sentence that the audit trail can be used as evidence in legal proceedings is recorded here as vendor wording and is deliberately not carried into the scope as a legal validity claim, and no identity assurance, qualified signature, eIDAS, compliance or timestamp claim is made. No signature verification feature is claimed, and the single verify mention elsewhere in the guide is not turned into one. Merge, compression, editing, conversion, OCR, protection and viewing are separate jobs and are not folded in, and the guide cloud storage steps are not carried, consistent with the settled scopes on this record. No limits claim is made because the Sign PDF section states none.

  218. 7a053ac

    The iLovePDF record, in the ocr job

    What it said
    The ocr job carried no scope, so it could be read as an accuracy guarantee, as offering language selection, as stating page or size limits, as converting to Office formats, as including handwriting, or as being documented in the user guide alongside the other tools.
    What it says now
    The job keeps ocr and gains the scope converting non selectable PDF files into selectable and searchable PDF with the iLovePDF web OCR PDF tool, where the OCR PDF tool page says it converts non selectable PDF files into selectable and searchable PDF, and where the official FAQ says that converting a scanned PDF to an editable document requires optical character recognition, described as a system that converts non selectable and scanned text to office documents, with no language options, accuracy figure, page or size limit stated on the pages read, from https://www.ilovepdf.com/ocr-pdf and https://www.ilovepdf.com/help/faq read on 2026-10-16 through the repository renderer and directly. The tool page states the one direction, non selectable PDF in and selectable and searchable PDF out. The FAQ states that scanned PDF conversion to an editable format requires OCR and describes OCR as converting non selectable and scanned text to office documents, adding that Scanned PDF to Office conversions have been possible since the last release. The user guide was searched and contains no OCR section at all: zero occurrences of OCR, searchable, recogn, language, scanned or text layer in its visible text, so no procedure, language option, limit or output statement could be drawn from it. The tool page phrase with high accuracy is marketing wording and is not carried as an accuracy claim, and no accuracy figure is stated. No language options or list are claimed, because none is documented. No page count, file size or account limit is claimed. The FAQ scanned PDF to Office statement is recorded here but not carried as an OCR output claim, because Office output is the separate convert job, and the OCR scope claims only the non selectable PDF to selectable and searchable PDF direction. No handwriting, extraction, viewing, editing, signing, protection, compression, cloud or PDF/A claim is made. The guide silence on OCR is recorded here so a later pass does not repeat the search expecting a section. The settled merge, compress, edit, convert and sign scopes are unchanged.

  219. 7a053ac

    The iLovePDF record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as offering owner passwords or permission flags, as promising a named encryption algorithm or strength, as a confidentiality or compliance guarantee, as covering transport encryption, as a numeric limit claim, or as including the neighbouring security and repair tools.
    What it says now
    The job keeps protect and gains the scope protecting PDF files with the iLovePDF web Protect PDF tool, where the official user guide says that to secure a PDF with a password you upload the document, choose a user password to encrypt and secure your PDF, and then hit the Protect PDF button to add the password to the file, that the separate Unlock PDF tool removes the password from a PDF file, and that each tool defers its limits to the account type, and where the Protect PDF tool page says it encrypts your PDF with a password, with no owner password, permission flags, algorithm or key length documented on the pages read, from https://www.ilovepdf.com/help/documentation and https://www.ilovepdf.com/protect-pdf read on 2026-10-16 through the repository renderer. The guide PDF SECURITY block documents Protect PDF with a user password and Unlock PDF which removes the password, and both sections defer limits to the account type. The tool page says the tool encrypts the PDF with a password and speaks of keeping sensitive data confidential. A case insensitive scan of the guide found no occurrence of permission, owner, AES, RC4 or 256, so no owner password, permission flags or algorithm strength is claimed. The tool page confidentiality phrase is marketing wording and is not carried as a guarantee, and no legal, compliance or security assurance is made. Transport level security wording elsewhere on the site, such as the SSL language in the FAQ, is deliberately not carried, because that protects the transfer rather than the PDF. No numeric limit is stated because the sections defer limits to the account type. Signing, merging, compression, editing, conversion, OCR and viewing are separate jobs and are not folded in, and the guide cloud storage steps are not carried, consistent with the other settled scopes on this record.

  220. 7a053ac

    The IronPDF record, in the primary edit job

    What it said
    The edit job carried no scope, so it could be read as general PDF content editing of images and layout, as including stamps, annotations, forms, DOM access, metadata, redaction or HTML authoring, as an OCR or conversion capability, or as a fidelity guarantee.
    What it says now
    The job keeps edit, remains primary, and gains the scope replacing existing text in PDF documents with the IronPDF .NET library, where the official Replace Text in PDF how-to says the replace text action can be applied to any PdfDocument object whether newly rendered or imported, using ReplaceTextOnAllPages with the old and the new text, ReplaceTextOnPage for a single page and ReplaceTextOnPages for multiple specified pages, with zero based page indexes, support for newline characters in the replacement text and an optional custom font and size on the page method, raising an exception when the old text cannot be found, and saving the edited PDF to a new file, with the page showing version 2026.10, from https://ironpdf.com/how-to/find-replace-text/ read on 2026-10-16 through the repository renderer. The page documents the three replacement methods and their scope, the zero based indexing, the newline support, the custom font and size parameter for the page method, the exception raised when the old text is not found, and a quickstart that loads an existing example.pdf, replaces text and saves to updated.pdf. The official Editing PDFs in C# tutorial lists a wider feature set including stamps, annotations, HTML footers, DOM access and metadata, which is deliberately not folded in, and neither are forms, page organisation, redaction or HTML authoring, since each is a separate operation or a separate job on this record. No image or layout editing of existing content is claimed and no fidelity or layout preservation guarantee is made, because the page states none. No OCR, conversion, extraction, merging, signing, protection, viewing, compression or accessibility claim is made, and the mention that redaction is a separate feature for permanent removal is not turned into an edit claim. The version context is the 2026.10 version and 2 August 2026 updated date shown on the page. The settled compress scope on this record is unchanged.

  221. 7a053ac

    The IronPDF record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage, as including PDF to Word or PDF to Excel, as OCR of scanned documents, as a fidelity or layout guarantee, as a PDF/A archival claim, or as a hosted service rather than an in-process library.
    What it says now
    The job keeps convert and gains the scope converting other formats into PDF and PDF pages into other formats with the IronPDF .NET library, where the official how-to guides document DOCX to PDF through the DocxToPdfRenderer class and RenderDocxAsPdf taking a file path, a byte array or a stream with DocxPdfRenderOptions for mail merge and page setup, image to PDF for single or multiple JPG, PNG or TIFF images through ImageToPdfConverter.ImageToPdf with customizable placement, and further official guides for HTML file, HTML string, HTML element, HTML zip, ASPX, JavaScript, Markdown, RTF, URL and XML to PDF, and document PDF to HTML through the SaveAsHtml method, PDF to SVG through the PdfToSvg namespace converting page by page, and rasterizing PDF pages into PNG, JPEG or TIFF image files through RasterizeToImageFiles, from the official guides at https://ironpdf.com/how-to/docx-to-pdf/, https://ironpdf.com/how-to/image-to-pdf/, https://ironpdf.com/how-to/pdf-to-html/, https://ironpdf.com/how-to/rasterize-pdf-to-images/ and https://ironpdf.com/how-to/pdf-to-svg/ read on 2026-10-16 through the repository renderer, with the remaining to PDF directions taken from the titles of official guides enumerated in the rendered documentation index at https://ironpdf.com/docs/. The DOCX guide documents the renderer class, the RenderDocxAsPdf input overloads and the DocxPdfRenderOptions mail merge and page setup settings. The image guide documents single or multiple JPG, PNG or TIFF inputs with placement options. The PDF to HTML guide documents the SaveAsHtml one line method. The rasterize guide documents PNG, JPEG and TIFF outputs through RasterizeToImageFiles. The PDF to SVG guide documents the PdfToSvg namespace and per page conversion through the Pages collection. The to PDF directions are named from official guide titles rather than from a generic convert page or the library name, so no universal format coverage is claimed. The PDF to HTML sentence about maintaining visual structure and layout is a vendor quality claim and is not carried as a fidelity guarantee, and no fidelity or lossless claim is made. PDF to Word and PDF to Excel are not claimed because no guide documents them. OCR is excluded because scanned recognition is the separate IronOCR product, and PDF/A archival export appears as its own how-to page and is not folded in. Editing, extraction, merging, signing, protection, viewing, compression and accessibility are separate jobs or features and are not folded in, and no cloud processing is claimed since this is an in-process .NET library.

  222. 7a053ac

    The IronPDF record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as an unqualified encryption strength or confidentiality guarantee, as a compliance certification, as claiming every reader enforces the permission flags, or as including password removal and signing rather than the documented password and permission settings.
    What it says now
    The job keeps protect and gains the scope protecting PDF documents with passwords and permissions using the IronPDF .NET library, where the official Passwords and Permissions guide documents a user password, also called the open password, which is required to open and view the document, and an owner password, also called the permissions password, which controls permissions for editing, printing and other actions, set through the SecuritySettings UserPassword and OwnerPassword properties, with permissions that can make a document unprintable, read-only and encrypted, an EncryptionType setting whose values are Rc4_128 for 128 bit RC4 as the default, Aes_128 for 128 bit AES and Aes_256 for 256 bit AES, each listed with the PDF version and minimum Acrobat reader it requires, a note that RC4 remains the default so existing code is unaffected but is cryptographically weak and that AES is recommended for new documents, and opening a protected document through PdfDocument FromFile with a password that the library detects as user or owner and uses to grant the corresponding access level, from https://ironpdf.com/how-to/pdf-permissions-passwords/ read on 2026-10-16 through the repository renderer. The guide defines both password kinds, says the user password is required to open and view the document, says the owner password grants full control over security settings, says settings can limit a document to be unprintable, read-only and encrypted, and gives the EncryptionType table with Rc4_128, Aes_128 and Aes_256 plus their PDF versions and minimum reader versions, together with the plain statement that RC4 remains the default but is cryptographically weak and AES is recommended for new documents. It also documents FromFile accepting an optional password and detecting whether it is a user or owner password. The algorithms are carried because the vendor names each one with its bit length and compatibility, so this reports documented fact rather than inferring strength, and the vendor weakness note is included rather than omitted. No compliance claim is made despite the enterprise compliance sentence, no confidentiality guarantee is made despite the confidential wording, and no claim is made that every reader enforces the permission flags, because the page does not state that. No password removal claim is made beyond the owner password controlling security settings. Signing is a separate job and is not folded in, and editing, conversion, extraction, OCR, merging, viewing, compression and accessibility are separate jobs. No transport security, legal assurance or cloud processing claim is made, and the licence and pricing attributes are unchanged.

  223. 7a053ac

    The IronPDF record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as a visual signature stamp only, as a certificate based signature only, as a compliance or legal validity guarantee, as PAdES or eIDAS certification, as a timestamping service, as a cloud signing service, or as including signature verification without saying so.
    What it says now
    The job keeps sign and gains the scope signing PDF documents with the IronPDF .NET library, where the official Signing PDF Documents guide distinguishes three separate things, applying a certificate based digital signature with an X509Certificate2 certificate, stamping a visual handwritten signature image onto a document, and adding an interactive form field for users to sign electronically, where the HSM guide documents signing through the PKCS#11 API with a UsbPkcs11HsmSigner built from the library path, a PIN, a token label and a key label so that private keys stay on the device, and where the Verify PDF Signatures guide documents reading verified signatures through PdfDocument GetVerifiedSignatures, which returns VerifiedSignature objects whose SignerCertificate and CertificateChain properties expose the signer X.509 identity, the issuing CA, validity dates, a SHA 256 thumbprint, the raw DER bytes and the intermediate certificates up to the root, from https://ironpdf.com/how-to/signing/, https://ironpdf.com/how-to/signing-pdf-with-hsm/ and https://ironpdf.com/how-to/verify-pdf-signatures/ read on 2026-10-16 through the repository renderer. The Signing guide separates the certificate based X509Certificate2 signature, the visual handwritten signature image stamp and the interactive signature form field. The HSM guide documents PKCS#11 signing with the UsbPkcs11HsmSigner constructor taking the library path, PIN, token label and key label, and signing with SignAndSave. The verification guide documents GetVerifiedSignatures and the SignerCertificate and CertificateChain properties exposing X.509 identity, issuing CA, validity dates, SHA-256 thumbprint, raw DER and intermediates. The three kinds are kept separate so a visual stamp is not presented as a certificate signature and the form field is named as its own operation. The vendor claims about authenticity, tamper detection, non-repudiation, bank level security, eIDAS and ESIGN compliance, audit trails and chain of trust are recorded here as vendor wording and are not carried as identity assurance, legal validity, qualified signature, compliance or trust guarantees, and no PAdES or eIDAS claim is made. Timestamps appear in the tutorial and HSM pages but were not read closely enough to quote, so no timestamp claim is included. No cloud signing service is claimed, and editing, conversion, extraction, OCR, merging, protection, viewing, compression and accessibility are separate jobs and are not folded in.

  224. 7a053ac

    The IronPDF record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as offering an ordering control, as preserving bookmarks or metadata, as lossless or layout preserving, as offering page range selection, as including split and page imposition operations, or as a cloud service.
    What it says now
    The job keeps merge and gains the scope merging PDF documents with the IronPDF .NET library, where the official How to Merge or Split PDFs guide says IronPDF merges multiple PDF files into one document using the Merge method, and its quickstart merges two documents loaded with PdfDocument FromFile into a single document that is saved with SaveAs, with the inputs documented as existing PDFs or PDFs rendered from an HTML string, file or URL, and with the guide stating no ordering, bookmark or metadata preservation behaviour for the merged output, from https://ironpdf.com/how-to/merge-or-split-pdfs/ read on 2026-10-16 through the repository renderer. The guide documents the Merge method for combining multipage PDFs and the CopyPage and CopyPages methods for splitting, shows a quickstart merging two documents loaded with FromFile and saving with SaveAs, and says inputs can be existing PDFs or PDFs created from an HTML string, file or URL. A search of the page found no MergePdf, InsertPage or RemovePage, and the only bookmark and metadata mentions are a cross reference to a separate Organize PDFs tutorial and a conditional merging scenario based on metadata criteria, neither of which states preservation. So the scope states no ordering, bookmark or metadata preservation behaviour, and makes no lossless or fidelity claim. Splitting and the CombinePages page imposition method are documented but are different operations and are not folded into the merge job. No page selection or range argument is claimed for merge because none is documented. Editing, conversion, extraction, OCR, signing, protection, viewing, compression and accessibility are separate jobs and are not folded in, and no cloud processing is claimed since this is an in-process .NET library.

  225. 7a053ac

    The IronPDF record, in the accessibility job

    What it said
    The accessibility job carried no scope, so it could be read as a Section 508, WCAG or ADA compliance guarantee, as certification or validation of output, as PDF/A archival conversion, as remediation of untagged scanned documents, or as general PDF generation.
    What it says now
    The job keeps accessibility and gains the scope exporting PDF documents to the PDF/UA accessibility standard with the IronPDF .NET library, where the official Export PDF/UA Documents guide says PDFs are exported using the SaveAsPdfUA method on a PdfDocument, or built with RenderHtmlAsPdfUA straight from HTML, that the library writes both PDF/UA-1 (ISO 14289-1) and PDF/UA-2 (ISO 14289-2:2024) while tagging the document so screen readers and other assistive technologies can read it, that PDF/UA defines how a PDF carries its structure through a tagged content tree, alternative text for images, a declared reading order and a document language, that an existing PDF can be loaded and converted to the standard in one call, and that examples produce PDF/UA-1 by default while PDF/UA-2 is written using a version parameter, from https://ironpdf.com/how-to/pdfua/ read on 2026-10-16 through the repository renderer, with https://ironpdf.com/tutorials/pdf-accessibility-csharp-pdfua-tutorial/ read for corroboration and https://ironpdf.com/how-to/pdfa/ read to confirm it is an archival export page. The guide documents the two methods, the two PDF/UA versions with their ISO numbers, the tagging purpose for screen readers and assistive technologies, the structural elements PDF/UA covers, the one call conversion of an existing PDF, and the default PDF/UA-1 with PDF/UA-2 through a version parameter. The vendor statements about Section 508 requirements and the tutorial language about WCAG, ADA, validation and remediation are recorded here as vendor wording and are not carried as compliance, certification, audit or remediation guarantees, and no claim is made that output passes a validator. PDF/A is excluded entirely: the PDF/A guide is an archival export page covering PDF/A-3b and PDF/A-4, and is not presented as accessibility. No OCR, no reading order correction for untaggable documents and no claim about scanned image only PDFs is made, because the pages do not state those. The tagging elements are carried as the guide own definition of what PDF/UA covers rather than as separately verified features. Editing, conversion, protection, signing, merging, OCR, extraction, viewing and compression are separate jobs and are not folded in, and no cloud processing is claimed.

  226. 7a053ac

    The JPedal record, in the primary edit job, retrieval result

    What it said
    The edit job carried no scope and nothing on this record recorded what the official documentation does or does not document about editing existing PDF content, so the gap could be mistaken for an unchecked candidate rather than a searched one.
    What it says now
    Still unresolved, and the exclusion is specific. Three first party IDRsolutions pages were read on 2026-10-16 through the repository headless renderer, each renderer exit 0. The JPedal product page at https://www.idrsolutions.com/jpedal/ describes the library as pure Java for rendering, extraction, viewing and manipulation, gives the current release as 2026.09, and shows a PdfManipulator example that loads a document, adds a page, adds text, adds an image, rotates a page, applies the manipulations and writes the document. The manipulation documentation index at https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/ states the minimum Java version as 17 with no third party dependencies and summarises the library as merging, splitting, redacting, signing, sanitizing and optimising. The Manipulate PDF files page at https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/pdf-manipulator, 23,042 characters of visible text, says JPedal provides a PDF Manipulator class for easy manipulation, editing, and creation of PDF files, and then lists its complete feature set: add a page, copy a page, remove a page, move a page, swap pages, reverse pages, reorder pages, rescale a page, rescale a page content, crop a page, rotate a page, add text, add an image, add a shape, add an annotation, embed a file, attach a file, isolate pages, n-up pages, add bookmarks, remove bookmarks, set the document info, get and set the document metadata, remove metadata, set the initial view, generate a table of contents, page and document queries, and sanitization operations including flatten layers, redact text, remove JavaScript, remove embedded files, remove hyperlinks, remove metadata, remove annotations, remove the initial view and remove blank pages. Every operation in that list either adds new content to a page, moves or changes page level structure, or removes content; none edits or replaces existing text, existing images or existing layout. The only editing vocabulary on the pages is the phrase manipulation, editing, and creation in the opening sentence, and a marketing summary, neither of which names an operation. So the documented surface is adding content, page organisation, redaction, metadata, signing, merging, splitting and sanitising, all of which are separate jobs on this record or separate capabilities, and none of which is editing existing PDF content. The job stays unresolved rather than scoped to adding text and images, because the adding operations are content insertion rather than editing existing content and the catalog job would then read as a claim that JPedal edits existing page content, which the documentation does not support. It is not removed because the vendor does not say editing existing content is unavailable; it simply documents no such operation. A page describing an operation that modifies existing text, images or layout, or an explicit vendor statement that this is not supported, would settle it.

  227. 7a053ac

    The JPedal record, in the view job

    What it said
    The view job carried no scope, so it could be read as including printing, the forms editor, annotations, rendering to image files, a servlet or hosted service, a rendering fidelity guarantee, or a claim about image formats being PDF content.
    What it says now
    The job keeps view and gains the scope viewing PDF files with the JPedal Java viewer, where the official documentation says JPedal includes a comprehensive PDF viewer that uses Swing and optionally JavaFX, that the Viewer is customizable and written in Java but can be run from other languages or the command line, that JavaFX is an optional plugin used to enable the page flow mode, and that a viewer is opened through the Viewer class with setupViewer and the OPENFILE command or by running java -jar jpedal.jar with the view option, with the product page listing single page and continuous page display, document navigation, text search, zoom, scale and rotate of pages, a customisable interface and look and feel, and image viewing capabilities, in a 100 percent Java viewer running on Java 17 and later, from https://www.idrsolutions.com/jpedal/java-pdf-viewer and https://www.idrsolutions.com/docs/jpedal/tutorials/viewer/view-pdf-in-java read on 2026-10-16 through the repository renderer. The documentation page documents the Swing and optional JavaFX viewer, the optional JavaFX plugin for page flow mode, the command line view option, and the Viewer class with setupViewer and OPENFILE. The product page documents the viewer features, including printing and a forms editor, and states the viewer runs as a Swing component in Java applications. The home page gives the runtime as Java 17 and above and the current release as 2026.09. Printing and the forms editor are documented but excluded, and so are annotations, editing, extraction, conversion, OCR, merging, signing and protection, each being a separate job or feature. Image viewing is carried only as the product page names it, so no claim is made that image formats are PDF content. No fidelity or rendering quality claim is made despite the comparison table marketing wording, and no browser, servlet or hosted service claim is made despite the any JVM and cloud hosted environments wording. The version context is the 2026.09 release shown on the site. The edit blocker recorded in the previous pass and all other attributes are unchanged.

  228. 7a053ac

    The JPedal record, in the extract job

    What it said
    The extract job carried no scope, so it could be read as OCR of scanned PDFs, as the hosted Text Extraction Web Service, as an accuracy or layout fidelity guarantee, as covering conversion to images or other document formats, or as including the viewer text search controls rather than the extraction API.
    What it says now
    The job keeps extract and gains the scope extracting content from PDF files with the JPedal Java library, where the official extraction documentation says JPedal is a pure Java PDF library for extracting text, images and metadata, and its capability table lists word list extraction returning each word with bounding box coordinates, unstructured text as seen on the page, structured text as a logical hierarchy of headings, paragraphs, tables and lists, text search returning the page coordinates of all matches for a search string, images in JPEG, JPEG2000, JBIG2 and PNG and other embedded formats, and metadata covering page count, dimensions, author, date, XMP and DocInfo fields, with extraction of unstructured text from a specified rectangular area, document outline extraction, and a structured text extractor whose file output modes include EPUB, HTML, JSON, Markdown, XML and YAML, on Java 17 or later with no third party dependencies, from https://www.idrsolutions.com/docs/jpedal/tutorials/extract/, https://www.idrsolutions.com/docs/jpedal/tutorials/extract-text/ and https://www.idrsolutions.com/jpedal/extract-structured-content read on 2026-10-16 through the repository renderer. The capability table names the six content kinds and their outputs, the page documents rectangular area extraction, the text index lists outline, unstructured, structured and Apache Tika text extraction plus search and encoding, and the structured content page shows the ExtractStructuredText API with output modes EPUB, HTML, JSON, Markdown, XML and YAML. The Text Extraction Web Service API is a hosted service and is deliberately not carried, and the Apache Tika path is recorded here but not claimed. No OCR claim is made, because the single OCR mention on the pages is not used to claim recognition of scanned images. No fidelity or accuracy claim is made, and no claim that extraction reproduces layout beyond the structured hierarchy the table names. Viewing, conversion, editing, merging, signing, protection, printing and accessibility are separate jobs or features and are not folded in, and text search is carried only as the extraction API capability rather than as a viewer control. The Java 17 floor and no third party dependencies are carried because the page states them. The edit blocker and all other attributes are unchanged.

  229. 7a053ac

    The JPedal record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage, as including conversion from images or Office files into PDF, as lossless or fidelity guaranteed, as PDF/A beyond PDF/A-3 level B, as an accessibility feature, or as a hosted conversion service.
    What it says now
    The job keeps convert and gains the scope converting PDF documents to other formats with the JPedal Java library, where the official Convert PDF to Image documentation says JPedal converts PDF pages to raster images in pure Java, writing output directly to AVIF, BMP, GIF, HEIC, JPEG, JPEG2000, PNG, TIFF or WEBP or returning a Java BufferedImage for in-memory processing, that all conversion methods accept either a float scaling value or a width and height integer array to constrain output dimensions and that the fit to size option preserves aspect ratio, and that the BufferedImage path creates no temporary files, and where the Convert PDF to PDF/A-3 documentation says the PdfA3Converter class converts PDF files into PDF/A-3 files from Java or the command line, that only conformance level B is currently supported, and that the conversion result lists any issues and any lossy changes made to fix them, with a minimum Java version of 17 and a legacy Java 8 build from August 2024 available, from https://www.idrsolutions.com/docs/jpedal/tutorials/convert-images/ and https://www.idrsolutions.com/docs/jpedal/tutorials/pdfa/convert-pdfs-to-pdfa-compliant-files read on 2026-10-16 through the repository renderer. The image page names the nine output formats, the BufferedImage in-memory path, the scaling and dimension options with aspect ratio preservation, and the no temporary files behaviour, and gives the Java 17 requirement with the legacy Java 8 build. The PDF/A page documents the PdfA3Converter class, the Java and command line paths, the level B only limitation, and the ConversionResult reporting issues and lossy changes. The formats are named exactly and no universal coverage is claimed, and no reverse direction from images or other files to PDF is claimed because only these two directions are documented on the pages read. The PDF/A claim is limited to PDF/A-3 level B and the page own lossy changes wording is carried rather than hidden, and PDF/A is not presented as accessibility. No fidelity or lossless claim is made for image conversion, and no web service or cloud claim is made. Viewing, extraction, editing, OCR, merging, signing, protection, printing and accessibility are separate jobs and are not folded in. The Java version statements are carried because the page states them. The edit blocker and all other attributes are unchanged.

  230. 7a053ac

    The JPedal record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as including timestamps or a timestamp authority, as signature verification, as a visible signature appearance, as identity assurance or legal validity, as a qualified or eIDAS signature, as a cloud signing service, or as a different signature level than the one documented.
    What it says now
    The job keeps sign and gains the scope digitally signing PDF files with the JPedal Java library, where the official Digitally Sign PDF files documentation, shown at version 2024.11, says JPedal supports PAdES Baseline Level B signatures and requires a PKCS #12 keystore containing the private key and X.509 certificate, a valid certificate, a certificate signature algorithm of SHA256 with RSA and a private key usage that includes digitalSignature, and documents the PdfSigner signPdf static method taking the input and output file locations, the keystore password, the keystore file, a signer name, a location, a reason and a document access permission level, where P1 allows no changes otherwise the signature is invalidated, P2 allows no changes except filling in forms, instantiating page templates and signing, and P3 includes those plus annotation creation, deletion and modification, and where the page contains no wording about timestamps, timestamp authorities, signature verification or a visible signature appearance, from https://www.idrsolutions.com/docs/jpedal/tutorials/digital-signatures/ read on 2026-10-16 through the repository renderer. The page states the PAdES Baseline Level B support, the PKCS #12 keystore requirement with private key and X.509 certificate, the valid certificate requirement, the SHA256 with RSA algorithm and the digitalSignature key usage, the signPdf parameters, and the three permission levels with the note that changes outside them invalidate the signature. A search of the page found no timestamp, TSA, verify, verification, visible, appearance, image, lossy or incremental wording, so no timestamp, timestamp authority, verification or visible signature claim is made. No identity assurance, legal validity, qualified signature, eIDAS or compliance claim is made, and no cloud signing service is claimed. Editing, conversion, extraction, OCR, merging, protection, viewing, compression and accessibility are separate jobs and are not folded in. The permission levels are carried as the page defines them, and the signature validity statement is about validity rather than a legal claim. The version context is the 2024.11 marker on the page, and a guessed product page URL returned the site 404 and is recorded here. The edit blocker and all other attributes are unchanged.

  231. 7a053ac

    The JPedal record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as offering owner passwords or permission flags, as promising a named encryption algorithm or key length, as a confidentiality or compliance guarantee, as transport security, as a hosted service, or as including password removal.
    What it says now
    The job keeps protect and gains the scope protecting PDF files with a password using the JPedal Java library, where the official Encrypt PDF Files in Java documentation, shown at version 2026.05, says PDF encryption restricts access to document content so that only recipients with the correct password can open or read a file, that JPedal provides an API to apply password protection and configure encryption settings programmatically through the PdfManipulator class, that a document is loaded and written with writeDocument taking the output document and a password which sets the password required to open the document, with byte array and OutputStream versions of the method, and that additional encryption behaviour is configured through EncryptionSettings before writing, the documented example being whether document metadata should be encrypted, while the page names no encryption algorithm, key length, owner password or permission flags, from https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/encrypt-pdf read on 2026-10-16 through the repository renderer, with https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/pdf-manipulator read to confirm the Encryption entry under settings and configuration. The page documents the PdfManipulator load and writeDocument flow, the password that sets the open password, the byte array and OutputStream overloads, and the EncryptionSettings object with the metadata encryption example. A search of the encrypt page found no algorithm name, key length, owner password or permission flag, so none is claimed, and the further options are recorded as deferred to an EncryptionSettings Javadoc that was not read. The access restriction sentence is carried as the vendor description rather than a confidentiality guarantee, and no legal or compliance assurance is made. Transport security, cloud processing, signing, editing, conversion, extraction, OCR, merging, viewing, compression and accessibility are not claimed and are separate jobs or features. The version context is the 2026.05 marker on the page. The edit blocker and all other attributes are unchanged.

  232. 7a053ac

    The JPedal record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as preserving bookmarks or metadata, as lossless, as offering page range selection, as including splitting or N-up imposition, as a cloud service, or as having an unstated ordering rule.
    What it says now
    The job keeps merge and gains the scope merging PDF files with the JPedal Java library, where the official Merge PDF files documentation, shown at version 2025.05, says JPedal merges PDFs into a single new file containing all the pages from the provided files and leaves the original files untouched, that the PdfMerge command line tool accepts input files and options in any order while merging the files in the order they appear in the arguments with the last argument always being the output file, and that the PdfMerge Java API offers static convenience methods for merging two files or an array of files into one, with overloads taking a password for each input and overloads taking a list of merge options such as the interleave option, while the merge page itself states nothing about bookmarks or metadata, and where the manipulation index separately says the merged output preserves fonts, images and page structure from each input file, from https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/merge-pdfs and https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/ read on 2026-10-16 through the repository renderer. The merge page states the single new file containing all pages, the untouched originals, the command line PdfMerge argument ordering rule and the output as last argument, the Java convenience methods for two files and arrays with password and option overloads, and the interleave merge option example. Ordering is carried because the page states it and not inferred from method names. The preservation statement is carried with its source attributed to the manipulation index and worded as the vendor words it, not as a lossless guarantee. No bookmark or metadata retention claim is made because the merge page has no such wording. No page range or selection claim is made, since the page says all pages are included and the only option named is interleave. Splitting, N-up and page manipulation are separate documented operations and are not folded in, and editing, conversion, extraction, OCR, signing, protection, viewing, compression and accessibility are separate jobs. Password-protected inputs are carried as documented overloads rather than as a protection feature, and no cloud processing is claimed. The version context is the 2025.05 marker. The edit blocker and all other attributes are unchanged.

  233. 7a053ac

    The JPGToPDF record, in the primary convert job

    What it said
    The convert job carried no scope, so it could be read as covering the other image and document converters on the same site, as reading text from images, as editing text inside the PDF, as a fidelity or lossless guarantee, as a cloud converter despite the record locality attribute, or as having a page limit.
    What it says now
    The job keeps convert, remains primary, and gains the scope converting images to PDF with the jpgtopdf browser tool at jpgtopdfmobile com, where the official page says it accepts JPG, JPEG and PNG inputs including mixed batches, produces one standard PDF with one image per page, lets you drag the images into the order you want so they are merged in exactly that order, offers page sizes of A4, Letter, Legal or fit to image, image quality of High, Medium or Low and file size of Larger sharpest, Balanced or Smallest file, runs the conversion in the browser on your device using the open source jsPDF library and never uploads the files, offers one tap ID document presets for Aadhaar, PAN card, passport and licence, states there is no watermark, no account and no page limit, and provides a Download PDF option and WhatsApp sharing, while the page says it does not read the text of the image and lists editing text inside the PDF as not supported, from https://jpgtopdfmobile.com/ read on 2026-10-16. The page documents the three input formats with mixed batches, one image per page output, drag to reorder with the order preserved, the four page sizes, the three image quality and three file size options, the on-device browser conversion using jsPDF, the four ID document presets, the absence of watermark account and page limit, and the download and WhatsApp share actions. It also states that the tool does not read the text of the image and that a searchable PDF needs OCR, and its comparison table lists editing text inside the PDF as not supported, so both are excluded by the vendor own words. Only the three named input formats are claimed, and the other converters in the site footer such as PNG SVG GIF TIFF PSD and BMP to PDF and the PDF to image tools are separate pages and separate tools, so none is folded in. Merge, compress, split, signing, protection, viewing, accessibility and editing remain separate jobs or separate site tools. The quality wording about High preserving the image as it came from the camera is recorded here as vendor wording rather than carried as a fidelity guarantee, and the quality and file size options are carried as the labels the page uses. No cloud processing is claimed because the page states the opposite. The page updated date of 30 September 2026 is recorded. A same sounding domain at jpgtopdf com was fetched first and not used because the record URL is jpgtopdfmobile com and the two pages describe different products.

  234. 7a053ac

    The LEADTOOLS PDF SDK record, in the primary edit job, retrieval result

    What it said
    The edit job carried no scope and nothing on this record recorded what the official LEADTOOLS PDF SDK surfaces do or do not document about editing existing PDF content, so the gap could be mistaken for an unchecked candidate rather than a searched one.
    What it says now
    Still unresolved, and the exclusion is specific. Three first party LEADTOOLS pages were read on 2026-10-16 through the repository headless renderer, each renderer exit 0. The PDF SDK product page at https://www.leadtools.com/sdk/pdf describes a PDF API for reading, writing, converting and viewing, with sections for viewer controls, PDF form fields, conversion, security and redaction, search and analysis, comparison, digital signatures, annotations and markups, compression with MRC, multipage processing, parse and extraction, portfolio and attachments, bookmarks, forms and PDF JavaScript. Its only editing claim is under Annotate and Mark Up PDF Files, which says LEADTOOLS supports reading, displaying, editing, and writing PDF annotations and markups, and the page says annotations let users comment and draw without making permanent changes. Its text handling claims are redaction and extraction, not editing. The Leadtools.Pdf help introduction at https://www.leadtools.com/help/sdk/dh/pdf/ says the assembly supports reading, viewing and saving PDF files and that developers can use the properties to create dynamic PDF viewing and editing applications, and then lists key features in which the only editing item is PDF Annotations, create, edit, delete, add and reply to annotations, with the annotation types enumerated. The same feature list includes PDF Document Text and Annotation Deletion, described as deletion of visible text, graphics and images with regions marked for deletion while preserving other document contents, which is redaction rather than editing, and it includes page operations such as extract, delete, insert or replace pages, which is page organisation. A search of that help page found no occurrence of edit text, edit the text, modify text, replace text, edit content, content edit, edit image or page content. The rich document editing that LEADTOOLS does offer, with text search, spell checking, undo and redo, copy, cut, paste, ruler, font styles and paragraph styles for PDF, DOC and DOCX, is documented for the Document Editor product at https://www.leadtools.com/sdk/document/document-editor-html5, which is a separate HTML5 and JavaScript product with a .NET or Java hosted back end rather than the PDF SDK this record names, so it is not evidence for this record edit job. The job therefore stays unresolved rather than scoped to annotation editing or page operations, because those are markup and page organisation rather than editing existing content, and the catalog job would then read as a claim that the PDF SDK edits existing page content, which these surfaces do not document. It is not removed because the vendor does not say content editing is unavailable in the PDF SDK. A PDF SDK help page documenting an operation that modifies existing text, images or layout would settle it, as would a decision that annotation editing satisfies the job.

  235. 7a053ac

    The LEADTOOLS PDF SDK record, in the convert job

    What it said
    The convert job carried no scope, so it could be read as universal format coverage from the product page headline figures, as a 100 percent accuracy guarantee, as OCR, as an accessibility or PDF/UA claim from the PDF/A mention, as covering every pair of the named formats, or as the whole SDK rather than one component.
    What it says now
    The job keeps convert and gains the scope converting documents to and from PDF with the LEADTOOLS Document Converter component, where the PDF SDK page says the Document Converter API can perform conversion to and from PDF and lists converting to and from PDF for PDF/A, other document and Office formats, images, HTML, EMF and SVG, together with conversion between PDF and PDF/A, conversion between PDF standards versions, linearizing PDFs for Fast Web View, distilling Postscript into PDF and converting PDF documents into ISO compliant PDF/A-1b files, and where the Document Converter page names Adobe Acrobat PDF and PDF/A, Microsoft Office DOC DOCX XLS XLSX PPT PPTX PST EML MSG and XPS, CAD formats such as DXF DWG and DWF, TIFF JPEG PNG EXIF and HEIC among raster images, Plain Text RTF HTML MOBI and ePUB, and IBM AFP MO DCA IOCA and PTOCA, says images and documents can be converted to SVG, and documents deployment on Windows, Linux, macOS, Android, iOS and Web for .NET 6 and later, .NET Framework, .NET MAUI, Xamarin, UWP, WinForms, ASP.NET and web services, programmed in C#, VB, XAML, Java and HTML or JavaScript, from https://www.leadtools.com/sdk/pdf, https://www.leadtools.com/help/sdk/dh/pdf/ and https://www.leadtools.com/sdk/document/document-converter read on 2026-10-16 through the repository renderer. The PDF SDK page gives the to and from PDF direction list and the PDF/A, standards version, linearization and Postscript distillation conversions, and the help feature list gives PDF/A-1B conversion and the Postscript Distiller. The Document Converter page gives the named format families, the SVG output statement and the deployment and language matrix. The headline figures of more than 165 file types and hundreds more raster formats are deliberately not carried, so no universal coverage is claimed and only named families are listed. The 100 percent accuracy statement is recorded here as vendor wording and is not carried as an accuracy or fidelity guarantee. The converter is named as a component of the Document Framework rather than the whole SDK, and OCR is not claimed even though the page says the converter uses OCR technology internally, because OCR is a separate job on this record. PDF/A is carried as an archival conversion target and not as accessibility. Editing, merging, signing, protection, viewing, compression and accessibility are separate jobs and are not folded in, no cloud conversion service is claimed, and the edit blocker is unchanged. The guessed help path that returned 404 is recorded here.

  236. 7a053ac

    The LEADTOOLS PDF SDK record, in the extract job

    What it said
    The extract job carried no scope, so it could be read as OCR of scanned PDFs, as updating metadata or form fields, as a viewer search feature, as an accuracy guarantee, as a cloud service, or as covering write operations because the source wording also mentions parse and update.
    What it says now
    The job keeps extract and gains the scope extracting content from PDF documents with the Leadtools.Pdf assembly in the LEADTOOLS PDF SDK, where the official help introduction says the assembly supports extraction of text, hyperlinks, annotations, form field data, digital signatures, bookmarks and metadata, and its PDF text and image extraction feature list names extraction, selection, searching and retrieving of text in PDF documents, detection of text links, indexing of text to an optimized speed, extraction of images embedded within a PDF document and saving them into many raster formats, extraction of internal links between pages, and creation extraction and update of document metadata covering title, author, subject, keywords, creation date, modification date, producer and PDF version, and where the PDF SDK page says data such as unicode text and form fields, images, fonts, annotations, shapes, hyperlinks, bookmarks, table of contents and document metadata can be parsed, updated and extracted, with PDF Portfolio schema, embedded PDF files and PDF attachments extractable, documented at help version 23.0.2024.2.29, from https://www.leadtools.com/help/sdk/dh/pdf/ and https://www.leadtools.com/sdk/pdf read on 2026-10-16 through the repository renderer. The help introduction and the extraction feature list name the five extraction capabilities and the metadata fields, and the PDF SDK page adds the broader parse and extraction data list plus portfolio, embedded file and attachment extraction. The source wording parse, update and extract is split so only the extraction portion is claimed, and metadata or form field updating and other write operations are not carried, because they belong to editing or other jobs. The searching and indexing wording is carried only as the extraction feature list states it and not as a viewer search control. No OCR claim is made, since the OCR engine is a separate job and the extraction features named are for text already in the PDF. No accuracy or fidelity claim is made and no cloud processing is claimed. Viewing, conversion, editing, merging, signing, protection, compression, accessibility and printing are separate jobs or features and are not folded in. The help version string is carried as the version context, and the edit blocker and settled convert scope are unchanged.

  237. 7a053ac

    The LEADTOOLS PDF SDK record, in the view job

    What it said
    The view job carried no scope, so it could be read as including zoom, rotate, thumbnails, bookmarks, outlines, page navigation or text selection controls, as annotation authoring, as conversion, composition, editing or comparison, as an accuracy or performance guarantee, or as printing and hosted use.
    What it says now
    The job keeps view and gains the scope viewing PDF files with the Document Viewer API framework of the LEADTOOLS PDF SDK, where the PDF SDK page says you load and view any PDF file using the Document Viewer API framework, and the Document Viewer SDK documentation describes an OEM ready document viewing solution for .NET in C# and VB and for HTML5 and JavaScript, with document viewing features including text search, memory efficient paging, inertial scrolling and vector display, that it can view raster and document formats alike, and that the HTML5 and JavaScript edition builds zero footprint web applications without the need for plug ins while the .NET edition targets desktop applications running on Windows and Xamarin, while the pages read document no zoom, rotate, thumbnail, bookmark, outline, page navigation or selection controls and the product page accuracy and speed wording is marketing rather than documented behaviour, from https://www.leadtools.com/sdk/pdf and https://www.leadtools.com/sdk/document/document-viewer read on 2026-10-16 through the repository renderer. The PDF SDK page attributes viewing to the Document Viewer API framework, and the Document Viewer page gives the viewing feature list, the raster and document formats statement, and the HTML5 and JavaScript and .NET edition descriptions. A search of the viewer page found zero occurrences of zoom, rotate, thumbnail, bookmark, outline, page navigation and selection, so none of those is claimed, and the only navigation occurrence is the site menu label. The annotation mention is not carried because the page does not say whether it means displaying or authoring annotations, and authoring belongs to a separate capability. The viewer page verb list of view, convert, compose, edit and compare is carried only as viewing. The PDF SDK marketing claims about impeccable viewing accuracy and speed topping market leading readers are recorded here as marketing and not carried as guarantees. Forms, printing, hosting and cloud use are not claimed, and extraction, conversion, editing, OCR, signing, protection, compression, accessibility and printing are separate jobs and are not folded in. The guessed help path that returned 404 is recorded here. The settled convert and extract scopes and the edit blocker are unchanged.

  238. 7a053ac

    The LEADTOOLS PDF SDK record, in the view job, correction of a false negative

    What it said
    The previous view scope stated that the pages read document no zoom, rotate, thumbnail, bookmark, outline, page navigation or selection controls. That statement was false for the PDF SDK page. The error arose because the earlier pass searched only the sibling Document Viewer page for those control terms and then wrote the conclusion as if it applied to all the pages read, including the PDF SDK page that had never been searched that way. The sweep tool description for this job carried the same false negative.
    What it says now
    The view scope is corrected. The official PDF SDK page at https://www.leadtools.com/sdk/pdf, re-rendered on 2026-10-16 with 11,565 characters of visible text, contains an Online Demo Applications section for the HTML5 and JavaScript Document Viewer stating that the following features are demonstrated: load a document from local disk and url, use thumbnail viewer for page selection, view any bookmarks included in the document, and interactive zooming and panning. Counts for those terms on that page are zoom 1, pan 2, thumbnail 2, bookmark 6, local disk 1 and URL 1. The scope now carries loading from local disk and url, the thumbnail viewer for page selection, viewing bookmarks included in the document, and interactive zooming and panning, together with the Document Viewer SDK page documented features of text search, memory efficient paging, inertial scrolling and vector display for .NET in C# and VB and for HTML5 and JavaScript, and records the .NET Windows and Xamarin desktop edition and the HTML5 and JavaScript zero footprint web edition. The page labels the list as features demonstrated, so the scope describes them as documented and demonstrated behaviour rather than tested SDK performance, and the PDF SDK page claims about viewing accuracy and speed are recorded as marketing rather than documented test results. The same demo list also names drawing annotations on the document and printing documents and annotations, and those are deliberately not carried in the view scope because annotation authoring and printing are separate capabilities and separate jobs; the sweep tool description was corrected to match. Rotate is not claimed and is recorded as absent because the PDF SDK page contains no occurrence of rotate, while the earlier entry had also listed it among undocumented controls without having searched that page. The settled convert and extract scopes and the edit blocker are unchanged.

  239. 7a053ac

    The LEADTOOLS PDF SDK record, in the ocr job, product boundary

    What it said
    The ocr job carried no scope and nothing on this record recorded that OCR is not documented as part of the PDF SDK assembly, so the job could be read as a supported PDF SDK capability and could be scoped later from the wrong product.
    What it says now
    Still unresolved, and the product boundary is now recorded. Four official surfaces were read on 2026-10-16 through the repository headless renderer or fetched directly. The PDF SDK page at https://www.leadtools.com/sdk/pdf, 11,565 characters of visible text, mentions OCR in only two ways: in the sentence that robust PDF controls to read, write, convert and view PDF files can be combined with state-of-the-art OCR, ICR, OMR, Forms Recognition, Virtual Printing and scanning APIs within LEADTOOLS to create complete solutions, and once in a demo application title in the example list, OCR Image to Image-over-text PDF in C#. It contains no OCR procedure, no OCR input or output statement, no OCR language list and no OCR API. The Leadtools.Pdf help introduction at https://www.leadtools.com/help/sdk/dh/pdf/, 4,581 characters with help version 23.0.2024.2.29, contains zero occurrences of OCR and zero of searchable, and its feature list covers signatures, annotations, forms, bookmarks, metadata, portfolio, PDF/A conversion, text and image extraction, page operations, encryption and optimisation, with nothing about recognising images. The Recognition engine page at https://www.leadtools.com/sdk/engine/recognition presents OCR as part of a separate Recognition SDK Technology alongside MICR, OMR, ICR, barcode and forms recognition libraries, not as a PDF SDK component. The page at https://www.leadtools.com/sdk/ocr now serves an Apryse OCR product page, reflecting the vendor rebrand, and describes OCR as an embedded, self hosted OCR SDK and an add-on to the Apryse Server SDK, with its own OCRModule API methods such as ProcessPDF and ImageToPDF, searchable PDF output with an invisible selectable text layer, structured text and coordinate output as JSON and XML, a language coverage section, and add on modules for structured output, so that surface belongs to the OCR product rather than to the LEADTOOLS PDF SDK this record names. Therefore the job is left unresolved rather than scoped: scoping it from the Apryse OCR page would inherit a sibling product capability, and scoping it from the PDF SDK page would rest only on a combination sentence and a demo title with no documented operation, input, output or limit. It is not removed because the PDF SDK page says the PDF controls can be combined with OCR within LEADTOOLS, so the capability exists in the family even though it is documented under a different component. A PDF SDK help page or tutorial documenting OCR use with the PDF assembly, including the image input types, the searchable PDF or text output and any language boundary, would settle it, as would a decision that the combined PDF and OCR demo is enough to satisfy the job.

  240. 7a053ac

    The LEADTOOLS PDF SDK record, in the sign job

    What it said
    The sign job carried no scope, so it could be read as including algorithms, digests, timestamps, signature appearance or validation policy, as a legal validity or non repudiation service, as an audit trail or trust chain, as cloud signing, or as simple electronic signatures being a PDF SDK native feature rather than Document Viewer support.
    What it says now
    The job keeps sign and gains the scope digitally signing PDF documents with the LEADTOOLS PDF SDK, where the PDF SDK page says the PDF SDK libraries can add digital signatures to PDF files, retrieve information from embedded signatures and verify signatures to see whether the PDF has been manipulated, and the Leadtools.Pdf help feature list says a digital signature can be added to a PDF document from a PFX file and that relevant information can be retrieved from the embedded signature, and where the same PDF SDK page says that when combined with the eSignature support integrated in the LEADTOOLS Document Viewer, developers can also add simple electronic signatures to the PDF before creating the digital verification certificate, with no algorithm, digest, timestamp, signature appearance or validation policy documented on the pages read, from https://www.leadtools.com/sdk/pdf and https://www.leadtools.com/help/sdk/dh/pdf/ read on 2026-10-16 through the repository renderer. The PDF SDK page gives the add, retrieve and verify behaviour and the Document Viewer eSignature attribution, and the help feature list gives the PFX file input and the retrieval behaviour. Certificate based digital signing is kept distinct from simple electronic signatures, and the electronic signature capability is carried with the attribution the page gives it rather than as a PDF SDK native feature. No algorithm, digest, key length, signature appearance, timestamp or validation policy is claimed, because the pages contain no such detail beyond PFX. The phrase much more secure is marketing and is not carried as a security or non repudiation guarantee, and verification is carried as documented behaviour of checking whether the PDF was manipulated rather than as a trust or legal claim. No legal validity, eIDAS, ESIGN, HIPAA, GDPR, audit trail, trust chain or cloud signing claim is made. Annotations, form fields, redaction, editing, conversion, extraction, OCR, protection, merge, compression, viewing and accessibility remain separate jobs. The help version string is recorded as the version context, and the settled convert, extract and view scopes and the edit and ocr blockers are unchanged.

  241. 7a053ac

    The LEADTOOLS PDF SDK record, in the protect job

    What it said
    The protect job carried no scope, so it could be read as including owner and user password behaviour, an encryption default or a granular permission matrix, as HIPAA or GDPR compliance, as a confidentiality or attack resistance guarantee, as password recovery, as redaction, or as transport security.
    What it says now
    The job keeps protect and gains the scope protecting PDF documents with encryption and permissions using the LEADTOOLS PDF SDK, where the Leadtools.Pdf help feature list, at help version 23.0.2024.2.29, documents PDF encryption and decryption as password encryption and decryption with 40 and 128 bit RC4 and 256 bit AES, setting or modifying document security permissions, and removing encryption for existing PDF documents, and where the PDF SDK page says the LEADTOOLS PDF libraries fully support PDF encryption and PDF permissions to control editing, printing at high and low resolution and copying content, with neither page documenting owner versus user passwords, encryption defaults or a permission matrix, from https://www.leadtools.com/help/sdk/dh/pdf/ and https://www.leadtools.com/sdk/pdf read on 2026-10-16 through the repository renderer. The help feature list gives the three encryption and decryption capabilities and the named algorithms and key lengths, and the PDF SDK page gives the permission categories of editing, printing high and low resolution and copying content. These pages name the algorithms, so they are carried exactly and no default algorithm is inferred, since neither page states one. The permission categories are carried exactly as worded and no matrix, granular flag list or owner versus user password behaviour is claimed, because the help page has zero owner and zero user password occurrences. The HIPAA and GDPR mention is recorded here as vendor framing and not carried as compliance, and no confidentiality, attack resistance, legal or non repudiation claim is made. Removing encryption for existing documents is carried because the help list documents it, but no password recovery or cracking is claimed. Redaction is handled through the Document Analyzer component on the same page and is excluded as a separate capability, and signing, editing, conversion, extraction, OCR, merge, compression, viewing and accessibility remain separate jobs. No transport security or cloud processing is claimed, and the help version is recorded. The settled convert, extract, view and sign scopes and the edit and ocr blockers are unchanged.

  242. 7a053ac

    The LEADTOOLS PDF SDK record, in the merge job

    What it said
    The merge job carried no scope, so it could be read as offering an ordering control or page selection for the merge, as preserving bookmarks, metadata, annotations or forms, as lossless or structure preserving, or as including the split, extract, delete, insert, replace, order manipulation and thumbnail operations documented on the same pages.
    What it says now
    The job keeps merge and gains the scope merging multiple PDF files into a single PDF document with the LEADTOOLS PDF SDK, where the Leadtools.Pdf help feature list, at help version 23.0.2024.2.29, has an entry for PDF file merging and extraction of pages which includes the operation merge multiple existing PDF files into a single PDF file, and where the PDF SDK page section on comprehensive multipage processing APIs lists merge multiple PDF files into a single file among its multipage features, while neither page states an ordering rule, a page selection option, or any bookmark, metadata, annotation or form preservation behaviour for the merged output, and the same pages list split, extract, delete, insert, replace, page order manipulation and thumbnail generation as separate multipage operations, from https://www.leadtools.com/help/sdk/dh/pdf/ and https://www.leadtools.com/sdk/pdf read on 2026-10-16 through the repository renderer. The help feature list names the merge operation and groups it with page operations and PDF version and encryption checks, and the PDF SDK page lists the merge among multipage features alongside split, page extract delete and insert, page order manipulation and thumbnail generation. No ordering rule and no page selection option is stated for the merge, so neither is claimed, and the separate page order manipulation feature is not folded into the merge job. No bookmark, metadata, annotation or form preservation claim is made, because no page states preservation behaviour for merged output, and those mentions belong to separate feature sections for bookmark creation and extraction, parse and extraction, and annotation editing. No lossless, fidelity, quality or structure preservation claim is made. Split, extract, delete, insert, replace and thumbnail generation are documented but separate operations and are not claimed here, and the version, encryption detection, re-encryption and pagination items are likewise separate. Conversion, extraction, editing, OCR, signing, protection, viewing, compression and accessibility are separate jobs and are not folded in, and no cloud processing is claimed. The help version is recorded as the version context, and the settled scopes and the edit and ocr blockers are unchanged.

  243. 7a053ac

    The LEADTOOLS PDF SDK record, in the compress job

    What it said
    The compress job carried no scope, so it could be read as a lossless or quality preserving reduction, as naming an algorithm or a size reduction percentage, as an AI accuracy or performance claim, as universal format coverage, as OCR of text-searchable files, or as including conversion and other jobs.
    What it says now
    The job keeps compress and gains the scope reducing PDF file size with the LEADTOOLS PDF SDK, where the PDF SDK page says LEADTOOLS uses artificial intelligence to compress and optimize both image and text-searchable PDF file types, that it maximizes image-based PDF compression with LEADTOOLS advanced MRC image segmentation and compression technologies, and that the PDF libraries can optimize text-searchable PDF files by analyzing document features to make the resulting file smaller and more efficient, and where the Leadtools.Pdf help feature list at help version 23.0.2024.2.29 documents a PDF Optimizer that allows optimization of image size, image compressions, and font embedding and sub-setting, with no compression algorithm, size reduction percentage, quality setting or lossless behaviour documented on the pages read, from https://www.leadtools.com/sdk/pdf and https://www.leadtools.com/help/sdk/dh/pdf/ read on 2026-10-16 through the repository renderer. The PDF SDK page gives the AI compression claim, the MRC image segmentation and compression technology for image-based PDFs, and the document feature analysis for text-searchable PDFs, and the help feature list gives the PDF Optimizer entry with image size, image compressions and font embedding and sub-setting. The vendor phrase about maintaining quality is recorded here as vendor wording and not carried as a visual quality guarantee. No algorithm, codec, percentage, target size, quality setting or lossless or lossy label is claimed, because a search found no lossless or lossy wording, no percentage and no named algorithm on the pages, and the only named technology is MRC image segmentation and compression which is carried as named. The artificial intelligence wording is carried because the page states it, without any AI accuracy claim. Text-searchable PDF files are carried as a file type the tool can optimize, not as an OCR capability, and OCR remains blocked separately. No conversion, extraction, editing, signing, protection, merge, viewing or accessibility claim is added, and no cloud processing is claimed. The help version is recorded as version context, and the settled scopes and blockers are unchanged.

  244. 7a053ac

    The CamScanner record, in the compress job, refreshed retrieval result

    What it said
    The earlier blocker recorded the compress route as returning an Error 404 page and rested on four surfaces. This pass found that the route now returns HTTP 200 while still serving the same not found content, and re-read the official skill file, so the blocker is refreshed and the soft 404 is recorded rather than mistaken for a working tool page.
    What it says now
    Still unresolved. Three checks were run on 2026-10-16. First, the route https://www.camscanner.com/compress-pdf now returns HTTP 200 with 5,645 bytes rather than the 404 status recorded earlier, and the variants compress_pdf and tools/compress-pdf return the same 5,645 bytes, but the rendered page is the site own Error 404 page, whose visible text reads Error 404, Oops, it seems that the web page you are trying to access is playing hide-and-seek with us, Back to home, plus the site footer. That is a soft 404: the status code changed but the content is still a not found page with no compress tool, no upload control, no settings and no instructions, so a status code check alone would now wrongly suggest a working page. Second, the official skill file at https://raw.githubusercontent.com/CS-CamScanner/camscanner/main/SKILL.md was re-fetched, HTTP 200 with 35,933 characters, and contains the word compress exactly once, in the error handling table, in the row for file size exceeds the maximum limit, whose cause is given as File exceeds 40 MB and whose handling is Compress the file and retry. That is advice to the user to compress externally before retrying, not a CamScanner compression command, and it also documents a 40 MB file size limit for the skill rather than any compression behaviour. The skill command table contains no compression command: zero command lines match a compress command, pdf compress is absent, and the only command lines in the file are auth status, image receipt, image merge-pdf and the generic invocation form. Third, the earlier pass had already read the twelve page official site sweep, the homepage toolbox with seventeen tools none of which compresses, the agent documentation capability index with no compression category and zero occurrences of compress, and the conversion capability page with no compress, file size or reduce wording; those findings are unchanged. The job therefore stays unresolved rather than scoped, and it is not removed because nothing on the surfaces contradicts compression, they simply do not document it. A first party page or CLI command that names a compression operation on an existing PDF or image, with its input, output and any size or quality control, would settle it.

  245. 7a053ac

    The CommonLook PDF record, in the edit job, refreshed retrieval result

    What it said
    The earlier blocker recorded the Remediator Tip and the Operations Tab operation list once. This pass re-checked the current manual so the blocker reflects the pages as they render now, and records the legacy URL redirect behaviour and the current manual location rather than only the earlier wording.
    What it says now
    Still unresolved, and re-confirmed against the current official manual on 2026-10-16. The CommonLook Manuals index at https://manuals.commonlook.com/ is reachable and links the CommonLook PDF User Guide, and the guide page at https://manuals.commonlook.com/resources/manuals/commonlook-pdf-global-access-user-guide/ still opens with the Remediator Tip: Using CommonLook PDF to fix your PDFs will not change the physical appearance of the document. The legacy URL at https://commonlook.com/user-guides/commonlook-pdf-user-guide/ does not serve the guide: the first response is HTTP 301 redirecting to http://allyant.com/user-guides/commonlook-pdf-user-guide/, which is what the previous entry recorded as a cross origin redirect whose destination serves its page not found response, and the material is published on the manuals host instead. The guide chapter list is the Toolbar and Ribbon, the Insert Tag and Convert Tag Tabs, the Settings Tab, the Windows Tab, the Operations Tab, the Tag Properties Panel and the Integrity Validator, which is a structure, tagging and validation surface. The Operations Tab page at https://manuals.commonlook.com/resources/manuals/commonlook-pdf-global-access-user-guide/commonlook-pdf-global-access-ribbon-tabs-operations/, 13,740 characters of visible text, lists operations that act on the tag tree, such as create tag, untag, flatten tags, merge tags, change tag order, split text, mark font as all caps, linearize table, insert row, insert LI, insert TOCI and insert LBody. A search of that page returns zero occurrences of edit text, modify text, replace text, edit content, edit image, page content and resize, and its thirteen occurrences of move all refer to moving tags. No operation is documented that changes existing text, images or layout, so the missing operation is any documented operation that modifies the rendered document content. Scoping edit would duplicate the settled accessibility job and would tell a reader the tool edits content, which the vendor own tip denies. The job is not removed because the vendor does not present content editing as unavailable as a product category, it states a boundary on the remediation workflow. A manual page describing an operation that changes the rendered document would settle it, as would a vendor statement that the tool does not edit content at all.

  246. 7a053ac

    The Documenso record, in the edit job, refreshed retrieval result

    What it said
    The earlier blocker read three user guide pages. This pass enumerated the entire current user guide, checked the field placement page, and re-read the upload and template pages, so the blocker records the full page inventory and the field placeholder boundary rather than three sample pages.
    What it says now
    Still unresolved, re-confirmed against the current official documentation on 2026-10-16. The documentation home at https://docs.documenso.com/ describes Documenso as the open source document signing platform, with a user guide, developer guide and self hosting guide, which matches the record tagline and its open source and self hosted attributes. The user guide index at https://docs.documenso.com/docs/users links twenty one pages, and their paths are the guide index, documents, add fields, add recipients, direct links, send, upload, getting started, create account, send first document, organisations, create team, members, overview, single sign on, settings, profile, security, templates, create template and use template. None of the twenty one pages is about changing document content, and there is no page for editing text, images, layout or page content. The upload page at https://docs.documenso.com/docs/users/documents/upload says documents are uploaded to prepare them for signing, that the supported formats are PDF and DOCX with a 50MB maximum configurable for self hosted instances, and that after processing the editor opens and the user is taken to the document editor to add recipients and fields. It also says a DOCX is converted to PDF on upload and warns that conversion can shift features, that form controls are flattened, and that encrypted and password protected PDFs are not supported. The add fields page at https://docs.documenso.com/docs/users/documents/add-fields says fields are placeholders that recipients fill in when signing, that you drag fields from the palette onto the document and position them, that you can drag a field to reposition it after placing, and it lists the available field types as signature, initials, name, email, date, text, number, checkbox, radio and dropdown, with the signature field capturing the recipient legally binding signature. The text field there is an input a recipient fills, not the document text. The create template page at https://docs.documenso.com/docs/users/templates/create describes the template editor as the place to configure recipients, fields and settings, and its Edit a Template section changes the template configuration rather than its content. Counting any of this as edit would duplicate the settled sign scope, which already covers upload, recipients and field placement, and would overstate the product as a PDF editor. The DOCX to PDF conversion on upload remains a conversion rather than an edit, and this record has no convert job. The missing operation is any documented operation that changes existing document text, images, layout, annotations or page content, in the web application or the API. A user guide page or API reference documenting such a change would settle it.

  247. 7a053ac

    The Foxit PDF Editor record, in the ocr job, resolution of the earlier blocker

    What it said
    The earlier blocker left the ocr job unresolved because no readable page described the OCR operation: the mobile manual had no OCR page, the product page marked OCR only as a yes row, and the desktop manual paths returned server errors. That was accurate at the time and is now superseded, because the documentation surface moved and the operation is documented.
    What it says now
    The job now has a scope. The help host changed: help.foxit.com and help.foxitsoftware.com both redirect to a knowledge base at https://kb.foxit.com/s/, whose search for ocr returns OCR articles, while the guessed desktop manual paths under help.foxit.com/manuals/pdf-editor/desktop/en_us/ now return HTTP 404 and the mobile manual at 11.2.0 still returns HTTP 200. Three articles were rendered on 2026-10-16. The main article at https://kb.foxit.com/s/articles/360058613291-Performing-OCR-on-a-scanned-document-or-image-PDF, headed Foxit PDF Editor for Windows and dated 14 August 2025, says running OCR on a scanned document or image based PDF makes the text searchable and editable, gives the scanner flow File Create From Scanner with Make Searchable (run OCR), gives the same flow under Convert From Scanner, describes dragging jpg tiff and png image files into the editor which converts them to PDF and prompts that some pages may contain unrecognized text, and for version 12 describes Convert then Recognize Text Current File with a dialog for the OCR language including a custom entry, the whole document or a page range, and output as plain text, searchable PDF or editable text, and it notes that version 11 and 12 support ended on 13 August 2025 with version 14 to upgrade to. The language pack article at https://kb.foxit.com/s/articles/13229978140564-Where-To-Download-OCR-Language-Packs, dated 2 July 2024, says the multilanguage version gets OCR language packs from the Foxit add ons page under OCR Language. The third party article at https://kb.foxit.com/s/articles/How-to-Use-Foxit-PDF-Editor-OCR-Feature-on-a-Document-Previously-Processed-by-Third-Party-OCR-Applications, dated 19 May 2025, records that PDFs already processed by third party OCR such as ABBYY FineReader may not be recognised as OCR-ed, that applying OCR again can create duplicate text layers and duplicate search results, and recommends the original non-OCR-ed document or re-imaging through the Foxit PDF Editor printer with Print as image before Convert then Recognize Text. The scope is limited to Windows because the articles are headed for Windows and the two mobile OCR articles in the same results are for Foxit PDF Editor Mobile and were not used. No accuracy, language count, handwriting, layout fidelity, cloud, privacy or compliance claim is carried, and the yes row on the product page remains a capability row rather than documentation of an operation.

  248. 7a053ac

    The ICEpdf record, in the sign job, refreshed retrieval result

    What it said
    The earlier blocker recorded a twelve page sweep of the official site with no signing vocabulary and a set of unreadable API reference paths. This pass re-checked the site, and also found that the GitHub organisation repository the record tagline relies on no longer exists, so the blocker is refreshed and that observation is recorded rather than left for a later pass to discover.
    What it says now
    Still unresolved. Six official pages were fetched again on 2026-10-16 with a desktop user agent and each returned HTTP 200: the product overview at 10,348 bytes, the PDF compliance page at 63,579 bytes, the documentation page at 6,576 bytes, the downloads page at 3,935 bytes, the demos page at 5,505 bytes and the supported platforms page at 7,146 bytes. The visible text of every one of them was searched case insensitively for signature, signed, signing, digital signature, certificate, PAdES, timestamp, time stamp and the bare word sign, and every page returned zero hits, including the PDF compliance page whose annotation and form field rows the earlier pass enumerated without finding a signature row. The home page at https://www.icepdf.org/ also returns HTTP 200 and describes ICEpdf as an open source Java PDF engine that can render, convert or extract PDF content, contributed and administered by ICEsoft Technologies, with source code and reference implementations available for download on the site; its rendered markup contains no GitHub link at all. That matters because the record tagline says the engine is maintained as a community project on GitHub, and the organisation repository https://github.com/ICEpdf/ICEpdf now returns HTTP 404 for both the web page and the GitHub API repository endpoint, as does the lowercase path. A GitHub repository search for icepdf returns only forks and mirrors, the largest being pcorless/icepdf with 124 stars, plus the svn2github mirror and others, and none of them is presented on the official site as the project home. So the GitHub surface is no longer available as a first party source for this record, and the claim in the tagline that the project is maintained on GitHub is now unsupported by the official site, which points at downloads on its own domain instead. That tagline question is recorded here for a separate candidate and was not changed in this pass, because this pass is the sign job only. The missing operation remains any documented first party signing, signature field, certificate handling or verification behaviour. A reachable first party developer guide or API reference, a product page naming a signature operation, or a replacement GitHub repository linked from the official site would settle it.

  249. 7a053ac

    The ICEpdf record, tagline and watchOutFor, in the repository claim

    What it said
    The tagline said the engine is maintained as a community project on GitHub, and watchOutFor said the community project is now maintained on GitHub, that it is where the current work is, and that the repository documents Apache-2.0 licensing, Java 11 build targets, current 7.4.1 artifacts and release notes covering editing and digital signing. The official site links no code host at all, and the GitHub organisation URL returns not found, so those repository facts were unsupported by any reachable official source.
    What it says now
    The tagline now reads that ICEpdf is an open-source Java PDF engine for rendering, converting and extracting, with no hosting claim. watchOutFor keeps the distinction between the legacy ICEsoft site, whose documentation page lists ICEpdf 4.2.2 release notes dated 23 August 2011 and whose footer reads a 2009 copyright, and the repository earlier revisions linked, records that the official site contains no link to any code host and that the GitHub organisation URL previously recorded returns not found, and withdraws the Apache-2.0, Java 11, 7.4.1 and release-notes facts as unverifiable rather than repeating them. The internal notes gain a dated update recording the withdrawal and the evidence. No replacement repository was invented, because the official site provides no authoritative one. The evidence, gathered on 2026-10-16: eight official pages fetched, home and product/overview.html and resources/documentation.html and downloads.html and demos.html and product/supported-platforms.html and sitemap.html returning HTTP 200 and news.html returning HTTP 404; absolute link harvest across those pages found zero links to github, gitlab, bitbucket, sourceforge or any other code host, with the only external hosts being www.icesoft.com at eight links, www.icefaces.org at nine, jira.icefaces.org at one and java.sun.com at one; the downloads page offers source through the site itself; the GitHub organisation URL family returns 404 for the web page and the API repository endpoint, and a repository search returns only forks and mirrors, the largest being pcorless/icepdf with 124 stars, none presented by the official site as the project home. The platform prose was checked and needed no change: the supported platforms page says the product is 100 percent Java and lists Windows, Linux and Solaris on Sun J2SE 1.5 to 1.7 and Mac OSX on Apple J2SE 1.5 and 1.6, matching the platform language already in the job scopes. The five job entries are untouched, the four settled scopes stand as documented from the official site at release 4.2, and the sign job remains unscoped and unresolved.

  250. 7a053ac

    The Level Access record, in the primary accessibility job

    What it said
    The accessibility job carried no scope and no ledger entry existed for this record at all, so the primary capability of a paid accessibility platform could be read as anything from a full WCAG conformance product to a consulting service, and coverage of documents, mobile apps and software was asserted only by the tagline.
    What it says now
    The job keeps accessibility, remains primary, and gains the scope auditing, testing, monitoring and remediating digital accessibility with the Level Access platform, where the official Audit and Test page says it combines automated scanning and monitoring with expert manual evaluation, that automated testing surfaces barriers in live, local, pre-production or secure environments with ongoing monitoring dashboards, that manual audits are run by experts including people with disabilities with unlimited validation testing, that the assets assessed range from websites and mobile apps to software, hardware and digital documents, and that its VPAT reports document conformance with the Web Content Accessibility Guidelines, Section 508 and EU standards, where the Automated Remediation and User Personalization page says the technology continuously scans the live site, finds common issues and automatically resolves them on every page load with suggested changes that can be reviewed, edited and approved, plus an optional personalization widget for colour contrast and saturation, font size and spacing, font selection, link highlighting, pausing animation, hiding images and voice navigation, and it states that complex issues cannot be detected through automation alone, where the Governance and Reporting page describes reporting and analytics dashboards with VPAT-inspired scores aligned with WCAG criteria and a Digital Accessibility Maturity Model, where the Mobile Testing page describes native iOS and Android rules with real device, simulator, emulator and cloud testing, CI or CD pipeline embedding and no code scans, and where the LevelDocs page describes guidance inside Microsoft Word for creating accessible Word files and exporting them to accessible PDFs with in-office remediation, read on 2026-10-16 from pages whose structured page data records a last modified date of 15 October 2026. Sources read: https://www.levelaccess.com/ for the product navigation, and the pages /audit-test/, /automated-remediation-and-user-personalization/, /governance-and-reporting/ and /mobile-testing/ fetched directly, and /leveldocs/ rendered in headless Chrome because a direct fetch returned a JavaScript challenge page. The structured page data records datePublished 2025-06-03 and dateModified 2026-10-15T07:20:52Z on the audit page and datePublished 2025-09-02 and dateModified 2026-10-15T07:22:10Z on the LevelDocs page. No WCAG, ADA, Section 508 or European Accessibility Act conformance is claimed for any capability, and the VPAT and the WCAG and PDF/UA compliant wording are carried as what the vendor says the reports or output document, not as an audit result. No coverage figure, issue detection rate, severity accuracy, AI accuracy, remediation guarantee or legal claim is made, and the vendor own statement that complex issues cannot be detected through automation alone is carried so the automation boundary travels with the capability. Consulting, Academy training and certifications, the SOC 2 trust centre, third party certification and accessibility statements are separate services and are not folded into the product scope. Document coverage is carried only as the pages describe it, LevelDocs guidance in Word and PDF export plus the audit page asset range, and not as a claim that arbitrary PDFs are remediated.

  251. 7a053ac

    The LibreOffice Draw record, in the primary edit job, retrieval result and blocker

    What it said
    The edit job carried no scope and no ledger entry existed for this job at all, so the tagline claim that the app opens PDFs for light editing could be read as documented PDF content editing, when the official material documents only a PDF import filter and PDF export.
    What it says now
    Left unresolved, and the distinction is now recorded. What the official sources do document: the LibreOffice Help conversion filter page at https://help.libreoffice.org/latest/en-US/text/shared/guide/convertfilters.html?DbPAR=SHARED, rendered in headless Chrome with 16,020 characters of visible text, has a section headed Filters for PDFIMPORT whose row reads PDF - Portable Document Format (Draw), API name draw_pdf_import, media type application/pdf (pdf), so a PDF import filter for Draw exists and is named. The same page and the Draw Guide document PDF export separately through the draw_pdf_Export filter. What the official sources do not document: no reachable page describes what happens when a PDF is imported into Draw or how imported content is edited. The current Draw Guide for LibreOffice 26.2 at https://books.libreoffice.org/en/DG262/DG262.html has fifteen chapters and none is about PDF import or editing, and both its chapter list and the earlier 7.2 chapter list contain no such chapter; its Print and Export chapter, DG26210-PrintExportEmail.html, documents exporting a document as PDF, Export Directly as PDF and the PDF Options dialog pages, and says nothing about opening or editing a PDF. The Getting Started Guide 26.2 Draw chapter, GS26207-GettingStartedWithDraw.html, has one PDF mention, that Draw files can also be exported in PDF format, with no import or editing wording. The official help source tree confirms the absence rather than a retrieval failure: the LibreOffice help repository directory source/text/sdraw/guide contains nineteen files, none with pdf or import in its name, and source/text/shared/guide contains one hundred and forty files of which the only pdf file is pdf_params.xhp, which is the PDF export options page. The Draw help module main page rendered with zero PDF occurrences. The product page at https://www.libreoffice.org/discover/draw/ rendered to the LibreOffice 26.8 suite page and confirms LibreOffice 26.8 is the current release, and the help pages identify themselves as LibreOffice 26.8 Help, while the current official guide books are at 26.2. Therefore the missing operation is any documented description of editing an imported PDF in Draw: which objects are editable, whether text or images or shapes can be changed, what is preserved or lost on import and save, and whether the result is saved as a Draw document or exported. The tagline claim that the app opens PDFs for light editing is not supported by any page read and is recorded here as a separate question rather than changed in this pass, which is the edit job only.

  252. 7a053ac

    The LibreOffice Draw record, in the editing claim and the preceding edit blocker

    What it said
    The previous blocker said no first party page describes what happens when a PDF is imported into Draw or how imported content is edited, and concluded that the tagline claim about opening PDFs for light editing was unsupported. That conclusion was wrong. The record own source field already named a Document Foundation wiki FAQ on Draw and PDF, and that FAQ describes the behaviour directly. The search that failed missed it because the help pages and the guide books were searched, and the wiki searches used phrases that the wiki search treated too narrowly and returned zero for.
    What it says now
    The tagline and the summary stand, and they are now backed by a named first party source. The Document Foundation wiki page Faq/Draw/118, How can I open a PDF in Draw, at https://wiki.documentfoundation.org/Faq/Draw/118, read through the wiki API on 2026-10-16 with a last revision date of 2026-09-04T06:35:22Z, says in its section Open and edit foreign PDF files: LibreOffice recognizes foreign PDF files as graphical documents, therefore PDFs are opened in Draw; the processing of foreign PDF files in Draw may be limited; the texts are usually no longer continuous texts, as you are used to for example in Writer; the texts and images are usually selectable as single snippets; these snippets are defined as Shape Text frames in the Draw navigator; these snippets can be edited; and in most cases this only makes sense if only single words or letters are to be exchanged. Its section Open and edit PDF files created in LibreOffice adds that an exported PDF file can then be edited in Draw or better yet in Writer. That is exactly the light editing wording the tagline uses, and it also supports the summary statements that imported PDFs open as drawing objects, that individual text snippets and images can be selected, and that imported pages behave as separate frames rather than flowing text, together with the watchOutFor note that long documents are tedious and layout can shift. The edit job therefore stays unresolved only because it has not yet been scoped from this source, not because the source is missing, and scoping it from the FAQ quotes above is the obvious next candidate. Two further points are recorded so they are not lost. First, the summary says X.509 digital signing, but neither the Draw Guide export chapter nor the export parameters help source contains X.509 or PKCS anywhere: they say a personal key or certificate is required, a certificate is obtained from a certification authority, and the parameters include SignatureCertificate and SignaturePassword, so X.509 is an unsupported specificity to qualify in a separate candidate rather than changed here. Second, PDF export itself remains documented as before, by the Draw Guide Print and Export chapter with its Digital signatures section and by the export parameters help source, so no export claim needed changing in this pass. No content was changed: the premise of the candidate, that the light editing language is stale, is not supported by the evidence.

  253. 7a053ac

    The LibreOffice Draw record, in the primary edit job, scoped from the FAQ

    What it said
    Two earlier entries need reconciling. The entry citing 6c878cf concluded that no first party page describes what happens when a PDF is imported into Draw or how imported content is edited, and left the edit job unresolved on that basis. That conclusion was false: the record own source field named a Document Foundation wiki FAQ on Draw and PDF, and that page describes the behaviour directly. The entry citing c75cbd9 withdrew the conclusion for the tagline but left the job unscoped. The false documentation-absence conclusion is now removed from the standing account, and what remains is this scoped job and an audit note recording why the source was missed.
    What it says now
    The edit job, which is primary, is scoped: importing and lightly editing PDF files with LibreOffice Draw, where the Document Foundation wiki FAQ How can I open a PDF in Draw at https://wiki.documentfoundation.org/Faq/Draw/118, revision 892580 dated 4 September 2026, says in its Open and edit foreign PDF files section that LibreOffice recognizes foreign PDF files as graphical documents so PDFs are opened in Draw, that the processing of foreign PDF files in Draw may be limited, that the texts are usually no longer continuous texts as they are in Writer, that the texts and images are usually selectable as single snippets, that these snippets are defined as Shape Text frames in the Draw navigator, that these snippets can be edited, and that in most cases this only makes sense if only single words or letters are to be exchanged, and its Open and edit PDF files created in LibreOffice section adds that an exported PDF file can then be edited in Draw or better yet in Writer, with no claim of continuous text editing, layout fidelity, support for every PDF, structure preservation, annotation or form editing, or automation. The revision metadata was read through the official wiki API on 2026-10-16: revid 892580, timestamp 2026-09-04T06:35:22Z, user Beluga, comment a text replacement of the community get involved link to the improve it page, which matches the revision Calum independently fetched. The vendor limitation is kept in the scope on the vendor own wording: processing may be limited, texts are usually no longer continuous, and editing only makes sense for single words or letters. No annotation, form, signing, protection or automation claim is folded in, and the earlier blocker observation that the help and guide books only document a PDF import filter and PDF export stands as a correct observation of those surfaces, but it was the wrong set of surfaces for the question. AUDIT NOTE ON THE MISSED SOURCE: the first pass searched the help pages, the Draw Guide 26.2 and the Getting Started Guide 26.2, and the help repository source listing, and then queried the wiki search API with multi word phrases such as PDF import Draw and editing PDF files LibreOffice, which returned zero because of how the wiki search treats multi term queries, so the wiki was treated as empty rather than searched page by page. The record own source field already named the FAQ, and reading the FAQ page by title through the API found it immediately. The lesson recorded here is that the cited sources on a record are part of the search space and must be opened before an absence is asserted.

  254. 7a053ac

    The LibreOffice Draw record, in the summary X.509 wording, withdrawal of an unsupported-specificity flag

    What it said
    The entry citing c75cbd9 recorded that the summary X.509 digital signing wording was an unsupported specificity, on the basis that neither the Draw Guide export chapter nor the export parameters help source contains X.509 or PKCS. That basis was true but incomplete, and the conclusion was false: the dedicated PDF digital signature help page documents X.509 certificates explicitly.
    What it says now
    No content change is needed, because X.509 is documented. The current LibreOffice 26.8 Help page Digital Signatures, at https://help.libreoffice.org/latest/en-US/text/shared/01/ref_pdf_export_digital_signature.html?DbPAR=SHARED, rendered on 2026-10-16 with 4,107 characters of visible text, says the signed PDF export uses the keys and X.509 certificates already stored in your default key store location or on a smartcard, that the key store can be selected under Tools Options LibreOffice Security Certificate Path, that a smartcard must already be configured for use by your key store, and it presents an X.509 Certificate control described as use this X.509 certificate to digitally sign PDF documents, allowing selection of an X.509 certificate to be used for signing the PDF export, with a Select Certificate dialog that displays all certificates in the selected key store and prompts for the key store password when protected. The same wording is present in the official help source file source/text/shared/01/ref_pdf_export_digital_signature.xhp, which contains four X.509 occurrences and no PKCS occurrence. The summary sentence therefore stands as recorded. What the earlier flag correctly observed is only that the Draw Guide Print and Export chapter and the export parameters help source do not name X.509 and speak instead of a personal key or certificate; that is a fact about those two surfaces, not about the product documentation as a whole. The dedicated digital signature help page is linked from the PDF export dialog and from the guide page link list, and it was not among the pages read when the flag was made. PKCS remains undocumented on the pages read and is not claimed anywhere in the record. No certificate type beyond X.509 is asserted, and nothing about signature standards, trust chains, legal validity, timestamping or verification is added here. The sign job on this record stays unscoped and is now a better candidate than before, because the digital signature help page documents the signing inputs, the key store route, smartcard support and the certificate selection dialog.

  255. 7a053ac

    The LibreOffice Draw record, in the sign job, scoped from the dedicated help page

    What it said
    The sign job carried no scope and no ledger entry for it existed on this record, so the unresolved state came from the inventory rather than from a recorded retrieval. The operation itself is documented: the dedicated Digital Signatures page for PDF export was never read in any earlier pass, which is the same wrong surface search that produced the FAQ miss on the edit job and the X.509 flag on the summary.
    What it says now
    The sign job is scoped: exporting a digitally signed PDF from LibreOffice Draw, where the current LibreOffice 26.8 Help page Digital Signatures at https://help.libreoffice.org/latest/en-US/text/shared/01/ref_pdf_export_digital_signature.html describes the Digital Signatures tab reached through File Export As Export as PDF, says the tab holds the options related to exporting to a digitally signed PDF, that the signed PDF export uses the keys and X.509 certificates already stored in the default key store location or on a smartcard, that a smartcard must already be configured for use by the key store, that Select opens a Select Certificate dialog displaying all certificates found in the selected key store and prompting for the key store password and for a smartcard PIN when protected, with the other fields accessible only after a certificate is selected, that the certificate password field takes the password protecting the private key associated with the selected certificate, usually the key store password or the smartcard PIN, that the optional Location, Contact information and Reason fields are embedded in the appropriate PDF fields, visible to anyone viewing the PDF, and may each be left blank, and that the Time Stamp Authority obtains a digitally signed RFC 3161 timestamp embedded in the signature so a viewer can verify when the document was signed, while if no TSA URL is selected, which the page gives as the default, the signature is not timestamped and uses the current time from the local computer, the operation being an export to a new signed PDF rather than signing the original document in place. The page was rendered on 2026-10-16 with 4,107 characters of visible text and read in full including the Time Stamp Authority paragraph and the related topics list, and the same wording is in the official help source file source/text/shared/01/ref_pdf_export_digital_signature.xhp. No trust chain validity, legal validity, non repudiation, eIDAS or ESIGN, audit trail or verification guarantee is claimed; the page statement about what signatures are used to ensure is carried as the vendor description of purpose, not as a guarantee that a viewer validates a chain. No PKCS detail and no certificate algorithm is claimed, because the page names none. Timestamping is carried exactly as worded, including RFC 3161 and the default no TSA case, with no claim about TSA trust or accuracy. No signing of the original in place, and no batch, automation or command line signing claim, is made even though the related topics list links the PDF export command line parameters page, which is a separate surface. AUDIT NOTE ON THE WRONG SURFACE SEARCH: the earlier passes on this record read the Draw Guide Print and Export chapter, the export parameters help source and the help and guide book surfaces, but never the dedicated digital signature help page that the guide page itself links in its related topics. That is the third absence or specificity judgement on this record that a good but incomplete surface set produced, after the FAQ named in the record source field and the X.509 wording in the summary; the corrective practice recorded across these entries is to follow the related and linked pages before concluding that a behaviour is undocumented.

  256. 7a053ac

    The LibreOffice Draw record, in the protect job

    What it said
    The protect job carried no scope and no ledger entry for it existed on this record, so the unresolved state came from the inventory rather than from a recorded retrieval, and the password and permission options that the export dialog documents were unrecorded.
    What it says now
    The protect job is scoped: exporting a password protected PDF from LibreOffice Draw, where the current LibreOffice 26.8 Help page Security at https://help.libreoffice.org/latest/en-US/text/shared/01/ref_pdf_export_security.html describes the Security tab reached through File Export As Export as PDF, says it specifies the security options of the exported PDF file, warns that document permission restrictions set by password are observed only by PDF readers compliant with version 1.5 of the format so in older readers the restrictions may have no effect, describes Set passwords opening a dialog where you enter the passwords with a password needed to view the PDF and an optional password allowing the person viewing the PDF to edit and or print, lists the Printing choices of not permitted, low resolution at 150 dpi with the note that not all PDF readers honor the setting, and high resolution, lists the Changes choices of not permitted, inserting deleting and rotating pages only, filling in form fields only, commenting and filling in form fields only, and any except extracting pages, and provides Enable copying of content and Enable text access for accessibility tools, where the PDF Export Command Line Parameters help source documents the corresponding parameter names EncryptFile, DocumentOpenPassword, RestrictPermissions, PermissionPassword, Printing, Changes, EnableCopyingOfContent and EnableTextAccessForAccessibilityTools with its type and default column giving boolean false for RestrictPermissions, long 2 for Printing, long 4 for Changes, boolean true for EnableCopyingOfContent and boolean true for EnableTextAccessForAccessibilityTools, and where neither page names an encryption algorithm, key size or encryption level. Sources read on 2026-10-16: the Security help page rendered with 2,950 characters of visible text and read in full including its related topics list, and the export parameters help source read through the official LibreOffice help repository with 13,400 characters of text. The parameters source describes DocumentOpenPassword as the password allowing a user to open the PDF when EncryptFile is true, RestrictPermissions as restricting some permissions that can be changed only when the user enters the correct password, and gives the Printing values 0 cannot print, 1 low resolution only, 2 maximum resolution and the Changes values 0 cannot be changed, 1 inserting deleting and rotating pages, 2 filling of form fields, 3 filling of form fields and commenting, 4 all changes except page extraction. No encryption algorithm, key size or encryption level is claimed, because a probe of the parameters source for RC4, AES, 128, 256, algorithm and encryption level returned zero occurrences and the security page names none. No default is inferred beyond the values recorded in the parameters type and default column. No cryptographic strength, password recovery, confidentiality, legal or compliance, or attack resistance claim is made. Protection is described only as options on the exported PDF file rather than protection of the original in place, signing and redaction are separate capabilities, and the accessibility tool flag is carried as named without becoming an accessibility conformance claim. This scope was made possible by reading the Security page linked in the export dialog and related topics, which is the same linked page practice recorded after the sign job, where the dedicated digital signature page had been missed.

  257. 7a053ac

    The LibreOffice Draw record, in the compress job

    What it said
    The compress job carried no scope and no ledger entry for it existed on this record, so the unresolved state came from the inventory rather than from a recorded retrieval, and the image compression options of the PDF export were unrecorded.
    What it says now
    The compress job is scoped: producing a smaller exported PDF by choosing the image compression options of the PDF export, where the current LibreOffice 26.8 Help page General at https://help.libreoffice.org/latest/en-US/text/shared/01/ref_pdf_export_general.html describes the Images section of the PDF export as setting the PDF export options for images inside the document, with Lossless compression selecting a lossless compression of images in which all pixels are preserved, JPEG compression selecting a JPEG compression level where a high quality level preserves almost all pixels and a low quality level loses some pixels and introduces artifacts but reduces file sizes, Quality entering the quality level for JPEG compression, and Change image resolution selecting to resample or down size the images to a lower number of pixels per inch with a target resolution selected for the images, where the PDF Export Command Line Parameters help source documents the corresponding parameters UseLosslessCompression described as specifying whether images are exported using a lossless compression format like PNG or compressed using JPEG with its type and default column giving boolean false, Quality described as the quality of the JPG export with a minimum inclusive value of 1 and a maximum inclusive value of 100 and a default of long 90, ReduceImageResolution described as reducing the resolution of each image to the value given by MaxImageResolution with a default of boolean false, and MaxImageResolution described as reducing all images to the given value in DPI with possible values of 75, 150, 300, 600 and 1200 and a default of long 300, and where the options apply to the exported PDF file. Sources read on 2026-10-16: the General help page rendered with 7,034 characters of visible text and read for the Images section, and the export parameters help source read through the official LibreOffice help repository with 13,400 characters. The page also notes that EPS images with embedded previews are exported only as previews and EPS images without embedded previews as empty placeholders, which is recorded here as a page observation rather than part of the compression claim. No size reduction percentage or factor is claimed, because probes of the parameters source for percent and percentage returned zero occurrences. No codec or format detail beyond the PNG and JPEG words the documentation itself uses is claimed, because JPEG2000, ZIP and codec all returned zero occurrences. The all pixels are preserved wording is carried as the page description of the named Lossless compression option, not as a guarantee for the exported document as a whole, and no visual quality, lossless, optimisation of an existing PDF, OCR or generic file size claim is made. The compression options are described as applying to the exported PDF file rather than to the original document, and no default is stated beyond the values recorded in the parameters type and default column. This is the fourth job on this record scoped from the linked export pages, after edit from the FAQ, sign from the digital signature page and protect from the Security page.

  258. 7a053ac

    The LibreOffice Draw record, in the convert job

    What it said
    The convert job carried no scope and no ledger entry for it existed on this record, so the unresolved state came from the inventory rather than from a recorded retrieval, and the documented Draw to PDF export paths, page and selection ranges, PDF version options, PDF/UA and hybrid PDF options and form and structure options were unrecorded.
    What it says now
    The convert job is scoped: exporting a Draw document as PDF, where the Draw Guide 26.2 chapter on printing and export, at https://books.libreoffice.org/en/DG262/DG26210-PrintExportEmail.html, describes exporting documents as a PDF file in its Exporting PDF format section, gives Export Directly as PDF on the Standard toolbar or File Export As Export Directly as PDF as the path that exports the complete document using default PDF settings with the note that page range, image compression and other printing options are not available that way, and gives Export as PDF as the path that opens the PDF Options dialog whose pages carry the options, and where the current LibreOffice 26.8 Help page General for PDF export describes the Range choices of All, Pages typed in the box using formats such as 3-6 and 7;9;11 and Selection, the PDF Version selector, Universal Accessibility PDF/UA creating a PDF that follows PDF/UA ISO 14289 with Tagged PDF automatically enabled and not disableable when PDF/UA is selected, Hybrid PDF embedding the ODF file so the exported .pdf contains both formats and remains fully editable in LibreOffice, the Forms option creating a PDF form whose submit format can be FDF, PDF, HTML or XML, and the Structure options of Tagged PDF, Export outlines, Comments as PDF annotations, Comments in margin and Export automatically inserted blank pages, and where the PDF Export Command Line Parameters help source documents SelectPdfVersion with the values 0 for PDF 1.7 as the default choice, 1 for PDF/A-1b, 2 for PDF/A-2b, 3 for PDF/A-3b, 15 for PDF 1.5, 16 for PDF 1.6 and 17 for PDF 1.7, together with PDFUACompliance, UseTaggedPDF, ExportFormFields, FormsType, ExportBookmarks and PageRange, and shows draw_pdf_Export in its command line examples, the documented direction being a Draw document exported to PDF rather than conversion of an existing PDF to another format. Sources read on 2026-10-16: the Draw Guide chapter read through the official books site with 30,405 characters of text, the General help page rendered with 7,034 characters, and the export parameters help source read through the official help repository with 13,400 characters. The direction is kept explicit: this is a Draw document exported to PDF, while the PDF import and light snippet editing behaviour is the separate edit job scoped from the Foundation FAQ. No conversion of an existing PDF into another format is claimed, no arbitrary format to format conversion is claimed, and PDF is the only output format asserted. No fidelity or preservation guarantee is made, and the hybrid PDF wording about remaining fully editable in LibreOffice is carried as the page description of that specific option. Signing, protection and compression live in separate tabs of the same dialog and in separate parameters on the same parameters page, and are recorded as the separate sign, protect and compress jobs on this record rather than folded into conversion. The PDF version value list is carried from the parameters source, which documents the values, while the General page only says the selector sets the PDF format, so no claim is made that the dialog presents the same list.

  259. 7a053ac

    The JPedal record, in the platforms versus prose candidate signal

    What it said
    The inventory carried an unresolved platforms versus prose candidate for JPedal, because the prose contains the word browser while platforms is only self_hosted. Nothing on the record explained that the word appears in an internal note about a claim that was previously removed, or that the vendor explicitly says JPedal cannot run in a browser, so the signal read as an unexamined platform contradiction.
    What it says now
    The signal is resolved with no content change, because the platform metadata and the prose are both correct. The only web or browser occurrence in the record prose is inside internalNotes, in the sentence that records the removal of an earlier wrong claim: the note says the claim that anything needing a browser-grade renderer is out of scope is removed. The tagline, summary, bestFor and watchOutFor contain no web, browser or online wording. The current official documentation answers the question directly. The page https://www.idrsolutions.com/docs/jpedal/faqs/running/can-jpedal-run-in-a-browser says Support for running Java in a browser has been removed, so you cannot run JPedal (or any Java code) inside a browser window, and adds that the separate BuildVu product converts PDF files into HTML5 so they can be viewed in a browser. The page https://www.idrsolutions.com/docs/jpedal/faqs/running/does-the-software-run-on-client-or-server says the vendor has customers using the software on both client and server, that the Viewer provides a viewer, that PdfDecoder provides Swing and non-Swing support, and that PdfDecoderServer can be used on the server in headless mode or for non-GUI and non-printing functions. The page https://www.idrsolutions.com/docs/jpedal/faqs/running/which-java-versions-does-jpedal-support says the minimum compatible version is Java 17, that JPedal is currently supported on Java 17, Java 21 and the latest Java version, that Java 17 support was announced in October 2025 to continue until at least October 2027, and that the latest version requires Java 17 or above with a link for customers to the last Java 8 release from August 2024. The product page https://www.idrsolutions.com/jpedal, rendered on 2026-10-16 with 14,214 characters of visible text and final URL with a trailing slash, says JPedal is a pure Java PDF library with no native binaries, no third-party dependencies and no JNI, that it runs on any JVM on-premise or in the cloud on Java 17+, and its comparison table names the embedded viewer as Yes (Swing) and list PDF to HTML conversion as a separate capability, and it states Current release: 2026.09. Against that, the record platforms array of self_hosted, the interfaces array of library, the tagline about a pure Java PDF library with rendered, extraction, viewing and manipulation, and the summary statements about no native dependencies, no JNI, running on any JVM from Java 17 and an embedded Swing viewer all stand. A Java library that runs inside the application on the customer infrastructure, including servers and cloud hosts, and that explicitly cannot run inside a browser window, is a self-hosted library, so no platform was added and none was removed. The viewer support documented for JavaFX and SWT, and the client and server deployment contexts, are deployment contexts of the same self-hosted Java library rather than a separate web platform, and they are recorded here rather than turned into platform entries. Fetch notes: support.idrsolutions.com now redirects to https://www.idrsolutions.com/docs/jpedal, and the requirements path under that documentation returned HTTP 404, so the Java and jar wording rests on the product page and the Java versions FAQ rather than a requirements page. Adding jpedal to the source backed platform set settles this signal in the inventory as verified-current, and the platform candidates document gains the row with the quoted support.

  260. 7a053ac

    The JPedal record, in the primary edit job, scoped from the manipulator documentation

    What it said
    The edit job was unresolved on the previous blocker, which had read the same operation list and concluded that the job could not be scoped to adding content and page operations because the catalog job would then read as a claim that JPedal edits existing page content. The observation was right but the disposition was too cautious: the vendor documents a manipulation and editing API whose operations are additions, page operations and removals, and that surface can carry a scope provided the existing content boundary travels with it.
    What it says now
    The edit job, which is primary, is scoped: manipulating PDF documents with the JPedal PdfManipulator class, where the official Manipulate PDF files in Java page at https://www.idrsolutions.com/docs/jpedal/tutorials/pdf-manipulation/pdf-manipulator, labelled v2026.07 in its heading, says JPedal provides a PDF Manipulator class for easy manipulation, editing, and creation of PDF files, documents loading a document from a File, byte array, InputStream or URL including password protected files, creating a new empty document, writing the document to a File, byte array or OutputStream, and applying manipulations that include adding text, images, shapes and annotations, adding, copying, removing, moving, swapping, reversing and reordering pages, rescaling pages and page content, cropping and rotating pages, embedding and attaching files, isolating and n-up pages, adding and removing bookmarks, setting document info, getting setting and removing metadata, setting the initial view, generating a table of contents, page and document queries, and sanitising or optimising operations that include flattening layers, redacting text, removing JavaScript, embedded files, hyperlinks, metadata, annotations, initial view and blank pages, rewriting streams, downsampling images and optimising colour palettes, with no operation among the sixty six documented headings that edits or replaces existing text, existing images or existing layout. The page was fetched on 2026-10-16 with 157,473 bytes of HTML and 29,361 characters of visible text, its complete heading list of 66 headings was read, and the load and write sections were read; the product page at https://www.idrsolutions.com/jpedal/ was re-read for its current release 2026.09, so the documentation labels itself v2026.07 while the product page states current release 2026.09. Probes of the page for edit text, modify text, update text, set text, change text, edit image, modify image and existing text returned zero occurrences outside the site navigation, whose only replace match is the unrelated page Replace the Viewer notification window. The record prose was checked against this and does not overclaim: the tagline says rendering, extraction, viewing and manipulation, and the summary says it handles manipulation, digital signing, forms including XFA, annotations and inspection, none of which asserts existing content editing. No content prose was changed, only the scope added, and the scope states the existing content boundary rather than inferring it from the words editing or manipulation or from the add text and add image operations. The prior blocker entry citing c12a90d remains a correct reading of the operation list and is superseded in disposition only.

  261. 7a053ac

    The LEADTOOLS PDF SDK record, in the primary edit job, scoped from the annotation and structural operations

    What it said
    The edit job was unresolved on the previous blocker, which recorded that the only editing claim on the product page was annotations and markups and that the help feature list named annotation create edit delete add and reply as the sole editing item, and concluded that the job could not be scoped because doing so would read as a claim that the PDF SDK edits existing page content. The reading was right and the disposition was too cautious: the vendor names an editing operation, annotations and markups, and that surface can carry a scope provided the existing content boundary travels with it.
    What it says now
    The edit job, which is primary, is scoped: editing PDF annotations and markup, and related structural updates, with the LEADTOOLS PDF SDK, where the official PDF SDK page says LEADTOOLS supports reading, displaying, editing, and writing PDF annotations and markups that work with Adobe Acrobat and other compliant readers, and that users can collaborate by writing comments and drawing shapes on top of the document without making permanent changes, and where the Leadtools.Pdf help feature list, Help Version 23.0.2024.2.29, documents PDF Annotations as load and save annotations, create, edit, delete, add and reply to annotations, show and hide all annotations, and read and write annotation viewing flags across the listed annotation types, and separately lists PDF Document Text and Annotation Deletion as deletion of visible text, graphics and images with regions marked for deletion, page operations of extract, delete, insert or replace pages, bookmark create, extract, update and delete, document metadata create, extract and update, form field loading, filling and saving, document initial view settings, and PDF JavaScript extract, write, replace and delete, with no operation on those pages that edits or replaces existing page text, images or layout, and with the rich document editing of font styles, paragraph styles and text formatting documented for the separate Document Editor product at https://www.leadtools.com/sdk/document/document-editor-html5 rather than for the PDF SDK. The pages were re-rendered on 2026-10-16: the product page at https://www.leadtools.com/sdk/pdf with 11,565 characters of visible text, the help introduction at https://www.leadtools.com/help/sdk/dh/pdf/ with 4,581 characters and Help Version 23.0.2024.2.29, and the separate Document Editor page with 4,315 characters. The product page other edit mentions are permissions described as controlling editing, and a comparison feature that reports text insertions, deletions and edits, neither of which is an editing operation. A search of the help page found no edit text, existing text or content edit wording. The annotation editing sentence itself says the annotations work without making permanent changes, which is the vendor own boundary on what that editing means. Redaction deletion, page operations, bookmark and metadata update, form field filling and JavaScript replace and delete are structural or removal operations and are carried as such rather than as page content editing. The record prose was checked and does not overclaim: the tagline names OCR, forms, conversion and automatic redaction alongside the PDF API, and the summary describes creating, loading, viewing and saving PDFs plus form fields, annotations, bookmarks, encryption and conversion. The prior blocker entry citing e67424d remains a correct reading of the same pages and is superseded in disposition only.

  262. 7a053ac

    The LEADTOOLS PDF SDK record, in the ocr job and the summary licensing claim

    What it said
    Two things. The ocr job remained unresolved with a blocker that predates the site OCR page changing, and the summary still asserted that text extraction, OCR, document analysis and automatic redaction are documented as part of the same product rather than as separate purchases, which contradicted the record own internal note from 2026-10-04 and asserted a licensing arrangement no page states.
    What it says now
    The ocr job stays unscoped and unresolved, now as a re-verified component boundary rather than an unread surface. Re-read on 2026-10-16: the PDF SDK page at https://www.leadtools.com/sdk/pdf, 11,565 characters of visible text, mentions OCR in exactly two places, the sentence that robust PDF controls to read, write, convert and view PDF files can be combined with state-of-the-art OCR, ICR, OMR, Forms Recognition, Virtual Printing and scanning APIs within LEADTOOLS, and one demo title in the example list, OCR Image to Image-over-text PDF in C#. It documents no OCR operation, input, output format, language or limit. The Leadtools.Pdf help introduction at https://www.leadtools.com/help/sdk/dh/pdf/, 4,581 characters at Help Version 23.0.2024.2.29, contains zero occurrences of OCR and zero of searchable, and no reachable OCR help assembly was found: /help/sdk/dh/Leadtools.Ocr/ returns 301 to /help/sdk/dh/leadtools.ocr/, which returns 404, and /help/sdk/dh/ocr/ returns 404 directly. The Recognition SDK Technology page at https://www.leadtools.com/sdk/engine/recognition presents OCR within a separate recognition technology family alongside MICR, OMR, ICR, barcode and forms recognition, and the Document SDK Technology page is another separate family. The page at https://www.leadtools.com/sdk/ocr now serves an Apryse OCR product page, describing embedded OCR that converts scanned PDFs into searchable PDFs and structured text with position data in over 80 languages, running inside the customer application as an add-on to the Apryse Server SDK on customer infrastructure, which is a different product and a different owner surface reached through this domain. Scoping the job from the combination sentence would turn a marketing link into a PDF SDK capability, and scoping it from the Apryse page would inherit a sibling product, so neither was done. The summary was corrected in the same content commit: it now says the PDF APIs are documented as combinable with OCR, ICR, OMR, forms recognition, virtual printing and scanning APIs within LEADTOOLS, that OCR and recognition are documented under separate LEADTOOLS SDK technology pages, that redaction is described with the separate Document Analyzer component, and that the site OCR page now resolves to an Apryse OCR add-on to the Apryse Server SDK, so which of a licence includes what needs confirming rather than assuming. A dated update was appended to the internal notes recording the re-check, the help path finding and the reason the separate-purchases wording was removed. The tagline phrase about OCR alongside the PDF API and the bestFor mention of recognition and PDF handling together were left as family level statements consistent with the combination sentence, because neither asserts that the PDF SDK performs OCR. The record claims array was checked and contains no OCR claim. What would settle the job: a PDF SDK help page or API reference documenting an OCR call, its input types, its output and any language boundary, or a vendor statement that OCR ships inside a PDF SDK licence.

  263. 7a053ac

    The LlamaParse record, in the primary extract job and the naming, output and interface prose

    What it said
    The extract job carried no scope, so the primary capability of a parsing product was unrecorded. The record prose also still described the product name as moving, named Markdown as the only output, described several parsing modes without naming the four documented tiers, said the vendor names four hard layouts, and listed only http as an interface although the vendor documents a CLI.
    What it says now
    The extract job, which is primary, is scoped to extracting content from documents with the Parse product on the LlamaParse platform, where the current official documentation at https://developers.llamaindex.ai/llamaparse/parse/ describes Parse as the document parsing product on LlamaParse, the platform from LlamaIndex, built for LLM pipelines, reading tables, charts, scanned pages and 130+ file types and returning clean markdown, text or JSON, documents four tiers fast, cost_effective, agentic and agentic_plus with a version pin per request, accepts a file_id or source_url with a required tier and version plus optional output_options, crop_box, page_ranges with max_pages and target_pages, user_metadata and webhook_configurations, returns per page and full markdown and text, a typed items tree of headings, text, tables, images and headers each with a bounding box, metadata, page images or screenshots and download URLs through the expand parameter, recovers tables with cell structure as markdown or HTML plus typed table items with rows, csv, html and md and an optional XLSX export, returns layout bounding boxes including granular boxes per word, line and cell, runs OCR automatically on pages with no text layer including scanned PDFs, phone photos and jpg, png, tiff, webp and heic images with a languages array of more than eighty language codes and camera photo correction, and parses selected pages only through page ranges, with the separate Extract, Classify, Split and Index products not claimed and the service hosted rather than self-hosted. The prose was corrected in the same content commit: the tagline now names Markdown, text or JSON; the summary names the four tiers and the version pin and lists the REST API, the Python, TypeScript, Go and Java SDKs, the CLI and the web UI as access routes; bestFor no longer claims four named hard layouts; the naming note now records the documented structure with Parse the parsing product on the LlamaParse platform and Extract, Classify, Split and Index as separate products; and interfaces is http and cli because the quickstart documents installing and running the llama-parse CLI. Sources read on 2026-10-16: the current documentation host reached after docs.cloud.llamaindex.ai returned 308 and docs.llamaindex.ai returned 301, with the pages read as raw Markdown, covering the overview, tiers, configuring-parse, retrieving-results, response-format, OCR and languages, tables, layout and bounding boxes, supported document types, getting started and quickstart; plus the product page https://www.llamaindex.ai/llamaparse and the pricing page https://www.llamaindex.ai/pricing. The curated claims were deliberately left as quoted because the product page still advertises best-in-class accuracy across 90+ formats and over 100+ languages, while the documentation says 130+ file types and more than eighty OCR language codes; that is a discrepancy inside the vendor own material, not a record error, and it is recorded here rather than resolved by picking one. No schema-based extraction claim was added, because that is the separate Extract product, and no self-hosted claim was made.

  264. 7a053ac

    The LlamaParse record, in the platforms and interfaces metadata, web platform added

    What it said
    The record carried platforms as api only and interfaces as http, while the current official getting started page documents a Web UI that a user works in a browser. The earlier commit in this pair introduced that documented Web UI wording into the summary, which caused the sweep to emit a fresh platforms versus prose candidate, and the metadata was the thing that was wrong rather than the prose.
    What it says now
    platforms is now web and api, ordered with web first as the other records in this index do, and the api edition platforms matches. interfaces is now http, cli and browser, because the same product is reached through the REST API, the documented llama-parse CLI, and the Web UI. The evidence is the getting started page at https://developers.llamaindex.ai/llamaparse/parse/getting_started/, read on 2026-10-16, whose description says the quick start covers document parsing using Python, TypeScript, Go, Java, the CLI, the REST API, or the Web UI, and whose Web UI section gives the steps to open the Parse app, optionally open Advanced Settings, upload a document or pick a sample use case, click Run Parse, and view the results directly in the browser, with a screenshot captioned as the Parse Web UI result view. A hosted application a user can operate in a browser is a web platform in this index vocabulary, and the same provenance supports the browser interface alongside http. The extract scope and the other prose corrections from the preceding commit were not changed here, the curated claims were still left as quoted despite the product page saying 90+ formats and 100+ languages while the documentation says 130+ file types and more than eighty OCR language codes, and the separate Extract, Classify, Split and Index products remain unclaimed.

  265. 7a053ac

    The LlamaParse record, in the edition and internal notes, alignment correction

    What it said
    The record contradicted itself after the preceding passes. The top level said platforms web and api and interfaces http, cli and browser, while internalNotes still said platforms api alone and interfaces http only, and the single edition read id api, name API, platforms web and api but interfaces browser only. That edition contradicted both the record level and its own name, and it would have shown a different story to anyone reading the editions data than the record told.
    What it says now
    The single edition is named Hosted service, with platforms web and api and interfaces http, cli and browser, matching the record level. The three stale internal-note paragraphs were rewritten. The naming paragraph now records the documented structure, with Parse the parsing product on LlamaParse the platform from LlamaIndex, and Extract, Classify, Split and Index as separate products reached through the same API key. The platforms paragraph now explains web and api from the REST API, the Python, TypeScript, Go and Java SDKs, the documented llama-parse CLI and the Web UI, keeps selfHostable for the documented enterprise hybrid cloud and VPC option, and keeps dataResidency unknown with the reason. The interfaces paragraph now explains http, cli and browser while keeping the distinction that these are entry points to a hosted service rather than an embeddable engine. A dated update was appended recording the alignment. The evidence is unchanged from the preceding passes and is still current: the documentation at https://developers.llamaindex.ai/llamaparse/parse/ documents Python, TypeScript, Go and Java SDK examples and a CLI tab, separates Parse from Extract, and its getting started page documents a Web UI workflow whose fourth step says to click Run Parse and view the results directly in the browser. After the change, searches of the record find no remaining api alone or http only phrasing outside the dated note that explains what was corrected, and the user-facing prose contains no claim that the name is moving beyond the watchOutFor sentence that says the naming has a documented structure rather than moving. The vendor discrepancy is preserved rather than harmonised: the product page still advertises best-in-class accuracy across 90+ formats and over 100+ languages, while the documentation says 130+ file types and more than eighty OCR language codes, and the curated claims remain quoted from the page. A scan of the other published records found twelve single-edition or multi-edition records whose edition interfaces are a subset of the record level, such as apryse local with library against a record level of library and browser, and those are the schema intended per-edition override rather than the same contradiction, so none was changed here.

  266. 7a053ac

    The Logrunlab PDF Compare record, in the primary compare job and the summary comparison claim

    What it said
    The compare job carried no scope, so the only comparison tool added last month had no recorded account of what its comparison means. The summary also still claimed it is the only tool in this index whose job is comparison, which contradicted both watchOutFor, which says other products can carry comparison without a Compare category, and the 2026-10-04 internal note that recorded the unique claim as removed.
    What it says now
    The compare job, which is primary, is scoped to comparing the text of two PDF files in a browser with PDF Compare, carrying the documented semantics and boundaries: extracts and normalizes text in the browser and shows mechanical Added, Removed and Changed results with page references; Changed is a paired block on both sides with differing wording, Added is text on the New side without a paired block on the Original side and Removed is the reverse; Original and New are positions chosen for the comparison so the labels are relative and the tool does not determine which file is historically newer or who made an edit; page references do not guarantee visual alignment or complete layout comparison and the labels do not indicate importance, risk or authorship; no meaningful text changes found does not prove the complete files are identical and the comparison excludes image-only or scanned content without a usable text layer, images and graphics, visual formatting or layout only changes, metadata and other non-text structure; it requires an existing usable text layer and runs no OCR, while a mixed PDF can be accepted if some pages contain extractable text and image-only pages are not compared; whitespace and some line wrap or page break extraction noise are normalized; each PDF is limited to 20 MB and 200 pages; password protected or encrypted PDFs are not supported in the current version; files, extracted text and results stay in the current browser session and are not stored as document content; TXT and MD are named as formats that may be added in future versions; and the linked contract version comparison workflow adds that Logrun Lab compares extracted text, does not read text in images or scans, does not compare page appearance and does not interpret legal meaning. The summary sentence claiming the only-in-this-index comparison position is removed, and the record now says only that it is the first record filed under the compare job, added on 2026-09-29. A dated internal note records the re-read of the unchanged 29,814 byte page, the route checks, the home page tool list and the workflow page being a guide for the same tool. Sources read on 2026-10-16: https://logrunlab.com/en/pdf-compare, https://logrunlab.com/en/compare-contract-versions and https://logrunlab.com/en, with route probes showing /en/pdf-compare/ returning 308 to /en/pdf-compare, logrunlab.com/ returning 308 to /en, and /en/privacy returning 200 while /privacy returns 404. The page footer reads 2026 Logrun Lab. The tagline, bestFor, watchOutFor, the five curated claims, platforms web, pricing free and the empty interfaces array, which is the index convention for a GUI application, all stand against the current page, and the 20 MB and 200 page limits and the recorded network test note were not changed.

  267. 7a053ac

    The Marker record, in the primary extract job and the hosted-service and mode prose

    What it said
    The extract job carried no scope, so the primary capability of a document conversion tool was unrecorded. Two summary statements were also wrong against the current README: the record said Marker is the engine behind Datalab commercial API, while the README now says the managed platform runs a version of the newer open source model Chandra with higher accuracy than Marker, and the modes were described as quality-first and CPU-oriented rather than by their documented names and device defaults.
    What it says now
    The extract job, which is primary, is scoped to extracting structured content from documents with the open-source Marker tool, where the README at https://github.com/datalab-to/marker says Marker converts documents to markdown, JSON, chunks and HTML, converts PDF, image, PPTX, DOCX, XLSX, HTML and EPUB files in all languages, formats tables, forms, equations, inline math, links, references and code blocks, extracts and saves images, removes headers, footers and other artifacts, works on GPU, CPU or MPS, and offers optional LLM assistance, with marker, marker_single and marker_gui entry points, balanced and fast modes that default by device, a disable_ocr option using the pure text layer that skips equations and scanned pages, a page_range option and converter classes for whole-PDF or table-only extraction, with surya-ocr 0.22.x as a declared dependency whose inference server the README says Surya auto-spawns using vllm on NVIDIA GPUs or llama.cpp llama-server on CPU and Apple Silicon, with hybrid use_llm mode documented as merging tables across pages, handling inline math, formatting tables and extracting values from forms across several LLM providers, and with the README limitations section saying very complex layouts with nested tables and forms may not work and forms may not be rendered well while noting the use_llm and force_ocr flags mostly solve those, and with Datalab managed platform recorded as a separate hosted product running Chandra rather than this open-source tool. The tagline now says the open-source tool ships alongside a separate Chandra-based hosted service, and the summary says the managed platform runs the newer Chandra model rather than that Marker is the engine behind the commercial API, and names the balanced and fast modes with their device defaults. A dated internal note records the current release check, the licence re-read, the pricing FAQ reconciliation and the retrieval limitation. Sources read on 2026-10-16: the README at 35,358 characters, MODEL_LICENSE at 14,656 characters, pyproject.toml, the GitHub repository metadata and latest release (v2.0.0, 2026-07-20), the PyPI JSON API (marker-pdf 2.0.0 uploaded 2026-07-20, Apache-2.0, requires Python >=3.10,<4), and https://www.datalab.to/pricing. MODEL_LICENSE is unchanged: commercial use is restricted above five million US dollars of prior-year gross revenue or raised equity or debt, except personal or research use, and entities offering a competing product or service are excluded, so watchOutFor stands. The pricing FAQ now explains the 2M versus 5M threshold difference as model-specific rather than contradictory, 2M for Chandra and 5M for Marker and Surya, and the pricing page still lists Free with 20 or 10 dollars monthly usage, Team at 400 dollars, rate limits of 25 and 400 and EU residency at 1.25x usage, while README also advertises 5 dollars in signup credits, which is a promotion alongside the allowance rather than a replacement. documentation.datalab.to returns 200 but the cited /limiting and /api-overview paths now return 404, so the earlier rate limit discrepancy between the pricing page and the documentation limits page could not be re-checked and is not restated. BUILD NOTE: the first build attempt after the edit failed with an Astro content schema error and an abnormal process exit because the first tagline draft exceeded the 160 character limit; the tagline was shortened to 153 characters and the rebuild, job scope, coverage, lastmod and full verify checks all passed.

  268. 7a053ac

    The Marker record, in the secondary convert job and the hosted-service wording

    What it said
    The convert job carried no scope, so the file-format conversion capability that sits under the primary extract job was unrecorded and the boundary between extracting content and converting to a representation was unstated. The hosted wording also needed a second correction: the previous pass replaced the claim that Marker is the engine behind Datalab commercial API with a claim that the managed platform runs Chandra, but the same README also says the platform API is powered by marker and surya, so the earlier replacement over-corrected in the other direction.
    What it says now
    The convert job, which is secondary, is scoped to converting document formats with the open-source Marker tool, carrying the documented output formats markdown, json, html and chunks with the output format, output directory, page range and paginate options; the input formats PDF, image, PPTX, DOCX, XLSX, HTML and EPUB in all languages; the PdfConverter for whole-document conversion and TableConverter for table-only conversion; image extraction and saving; and the marker, marker_single, marker_gui and Modal deployment entry points, with the boundary that the documented outputs are extracted representations of the source rather than a converted PDF or Office document, so no PDF-to-PDF or Office round-trip conversion is claimed. The summary now carries both README statements about hosting: Datalab documents its platform API as powered by marker and surya, and separately advertises a managed platform running the newer Chandra model, so the open-source tool and the hosted products overlap without being the same model everywhere, and the open-source package and hosted services are licensed differently. The tagline says the open-source tool ships with a hosted API and managed platform, at 147 characters within the 160 character schema limit. A dated addendum records the over-correction and names the two README sections it came from. Sources read on 2026-10-16: the README at 35,358 characters on master, with the current release unchanged at marker-pdf 2.0.0 from PyPI dated 2026-07-20, the GitHub latest release v2.0.0, and the repository last pushed 2026-10-14. The platforms and interfaces arrays, the extract and ocr jobs, the model licence warning with its five million dollar thresholds and competing-product clause, the qualityScore values and the pricing wording were left unchanged. Extraction remains the primary job and the two are kept distinct: extract is the reading of content, convert is the choice of output representation.

  269. 7a053ac

    The Marker record, in the secondary ocr job

    What it said
    The ocr job carried no scope, so the OCR path of a tool whose README says layout, OCR and table recognition all run through the surya vision-language model was unrecorded, and the distinction between the Surya project that Marker depends on and Marker itself was only in the internal notes.
    What it says now
    The ocr job, which is secondary at qualityScore 78, is scoped to running OCR with the open-source Marker tool, covering the local Surya VLM inference server spawned on first use with vLLM in Docker on NVIDIA GPUs and llama.cpp elsewhere and addressable through SURYA_INFERENCE_URL, the VLM being used for OCR in both balanced and fast modes and for layout in balanced mode; balanced re-OCR of the whole page when embedded text is bad, fast mode limiting model use to equations, surgical block-level repair of garbled or empty blocks and a single full-page pass for scanned or mostly bad pages so a clean digital document without equations never starts the server; tables reconstructed from the text layer with scanned tables from full-page OCR and low-confidence reconstructions falling back to the model; force_ocr, strip_existing_ocr and disable_ocr; OCR-only extraction through OCRConverter at marker.converters.ocr via Python or marker_single with the converter class, and keep_chars keeping characters and bounding boxes for digital PDFs while model-processed pages return block-level HTML without character boxes; and the Surya boundary, with Surya a separate Apache-2.0 project at https://github.com/datalab-to/surya distributed as the surya-ocr package at 0.22.1 on PyPI and pinned at 0.22.x by marker, describing itself as a 650M parameter OCR model with a claim of an internal benchmark of 91 languages and model weights under the model licence rather than the code licence. No conventional third-party OCR engine, no hosted OCR service and no accuracy guarantee is claimed, and the language claim is carried as Surya own benchmark claim rather than as a Marker guarantee. Sources read on 2026-10-16: the Marker README at 35,358 characters on master, the repository converter listing confirming marker/converters/ocr.py alongside pdf.py and table.py, the Surya repository metadata and README at 25,954 characters with default branch master and Apache-2.0 code, pushed 2026-09-11, the Surya PyPI JSON API showing surya-ocr 0.22.1 uploaded 2026-07-20 with Apache-2.0 and Python >=3.10,<4, and the marker pyproject pinning surya-ocr >=0.22.1,<0.23.0. Marker current release remains marker-pdf 2.0.0 from 2026-07-20. The tagline, summary, bestFor, watchOutFor, platforms and interfaces were already consistent after the extract and convert corrections and were left unchanged, and the hosted boundary remains recorded in the summary as Datalab own hosted products rather than as a Marker OCR service.

  270. 7a053ac

    The Master PDF Editor record, in the edit, merge, protect, convert, sign, ocr and compress jobs, plus the redaction verdict and batch support

    What it said
    Seven jobs on this record carried no scope, so the documented editing, merging, protection, conversion, signing, OCR and compression behaviour was unrecorded. Two fields also contradicted the sources: redaction was unverified while the manual states permanent removal and the record notes said removes_content, and batchSupported was null while the manual documents batch signing of multiple documents.
    What it says now
    All seven remaining jobs are scoped from the online manual read on 2026-10-16, while the existing view scope is preserved. edit is scoped to editing existing text objects by selecting and double-clicking, inserting simple and formatted text, changing font name, size, colour and style in the Object Inspector, selecting and modifying images with SVG, TIFF, JPEG, BMP, PNG and PPM insertion, moving and resizing objects, editing object properties, creating, editing and filling interactive forms, with the manual limits that per-word attributes require splitting the text object and that editing can fail when the font lacks characters because only used characters are embedded, and with no reflow, layout preservation or scanned-page text replacement claimed. merge covers Document Insert Pages with Add Files and Add Folder, Up and Down ordering and per-document page ranges, Extract Pages options for a single PDF, page ranges or multiple files with bookmark export and delete-after-extracting, splitting by page count or file size, and creating a PDF from PDF, SVG, SVGZ, TIF, TIFF, PNG, JPG, JPEG, BMP and PPM with ordering, page selection and insertion position. protect covers the four security instruments, password encryption with open and permission passwords through File Properties Security, certificate encryption for a list of valid recipients, and notes redaction and signing as separate jobs, with no algorithm, key length or owner versus user password detail claimed. convert covers export to images in BMP, JPEG, PNG and TIFF with page range, per-page naming, JPEG quality and TIFF compression, export to Word with page range and fill vertical space, export to TXT, CSV and Excel, Convert to PDF/A at levels 1a, 1b, 2a, 2b, 2u, 3a, 3b and 3u, creation of a new PDF from PDF, SVG, SVGZ, TIF, TIFF, PNG, JPG, JPEG, BMP and PPM files, and the Windows-only virtual PDF printer, one way with no round-trip Office editing claim. sign covers certificate-based signature checking, creating and signing, invisible signature fields, batch signing, the statement that the vendor provides no certificates, and the explicit warning that initials are not a digital signature and do not guarantee authenticity. ocr covers Document OCR converting image text to editable searchable PDF text with page range and language selection, tesseract traineddata language installation and the minimal-language recommendation. compress covers File Save Optimized As with remove unused elements, flatten form fields and merge layers, and Document Optimize Scanned Pages with JPEG or ZIP for colour and grayscale and ZIP or CCITT Group methods for black and white, with the vendor warning about image damage and print quality. The redaction verdict is now removes_content on a documentation basis with the exact manual sentence, a 2026-10-16 read date and a caveat that the removal is documented rather than independently tested, watchOutFor no longer calls the field unresolved, batchSupported is true, and the summary names the export formats beside PDF/A. Version and pricing were not asserted: the what-s-new page still shows Version 3.7.10 dated 19 May 2016 while the product page calls the line Master PDF Editor 5, and the purchase page served no price or currency, so the previously recorded USD 79.95 figure and its GBP conversion were left unchanged. The local edition library interface follows the same in-process convention as other desktop records with the record-level interfaces empty, so no edition change was made.

  271. 7a053ac

    The CamScanner record, in the secondary compress job

    What it said
    The compress job was left unresolved by three successive blockers which concluded that no compression capability was documented. Those blockers were accurate about the surfaces they read, the Play listing, the homepage and its seventeen-tool toolbox, the /compress-pdf route rendering the site error page, the skill file and the agent documentation, but none of them read the FAQ, which is where the capability is documented and limited.
    What it says now
    The compress job is scoped from the FAQ entry How do I reduce the size of a saved document at https://www.camscanner.com/faq, rendered on 2026-10-16 with 31,626 characters of visible text. The entry says CamScanner does not support compressing a document to a specific size, and documents the ways size can be reduced: adjusting the crop area, where the smaller the crop the smaller the image; changing the image filter, where black and white or Ink Saver reduce image size; sharing as images, where after Share as Image files over 5 MB can be shared at Actual Size or compressed proportionally; and sharing as a PDF, where Edit PDF at the bottom of the document screen then Reduce Size on the preview screen reduces it, with very large documents advised to share as a link instead. The scope carries those paths and the vendor limit, and states that this is size reduction of CamScanner documents through scanning choices and share or preview actions rather than a compressor with a target size, quality level, resolution setting, algorithm or percentage, none of which is claimed. The summary now says document size reduction through crop area and filter choice plus a Reduce Size share action rather than naming compression as a feature, and watchOutFor now carries the vendor statement that compression is not to a specific size. Supporting negative evidence recorded in the same pass: the English sitemap at https://www.camscanner.com/__sitemap__/en.xml lists 23 routes and none is a compression, reduce or optimisation page; the agent documentation sitemap lists 105 pages with no compression route; and the routes /compress, /reduce-pdf-size and /pdf-compress are not in the sitemap and produced no rendered content, so the Reduce Size control is an in-app and share-flow action rather than a web toolbox tool. The previous compress blockers are superseded on the FAQ evidence rather than deleted, and their surfaces remain a correct record of what was checked. The other jobs on the record, including the convert, edit, sign, protect and ocr scopes, and the owner, pricing, platforms and interfaces fields were left unchanged.

  272. 7a053ac

    The CommonLook PDF record, in the edit job, refreshed retrieval result and blocker

    What it said
    The edit job was unresolved on three earlier blockers that recorded the absence of content editing. This pass found new first party knowledge base articles titled Find and Replace, which could have read as content replacement, and a search-indexed legacy page titled Replace Text, so the blocker needed re-checking against those rather than being carried forward on the earlier surfaces.
    What it says now
    Still unresolved, and now checked against current product, manual and knowledge base sources read on 2026-10-16. The product page at https://allyant.com/pdf-accessibility-software-solutions/commonlook-pdf/ returned 144,690 bytes and 8,780 characters and describes AI tagging, advanced or simplified remediation editors and a four step upload, select editing, remediate and download workflow, with every use of editing referring to the remediation editor. The current manual guide still carries the Remediator Tip that using CommonLook PDF to fix your PDFs will not change the physical appearance of the document, and its chapters are toolbar and ribbon, insert and convert tag tabs, settings, windows, operations, tag properties and integrity validator, with no content editing chapter. The Operations Tab page at 13,792 characters says it offers tag creation and editing operations and contains zero occurrences of replace. Tag Properties documents ID, type, title, alt text, actual text and expansion text, which are accessibility properties. The Allyant knowledge base Find and Replace articles, one for the desktop version modified 11 September and one for the web advanced editor modified 14 September, say the tool locates repeated content and applies Alt text or Actual text to every matching instance at once by placing symbols in Span tags and confirming them in the Affected Elements panel, so it assigns accessibility text to tags rather than replacing visible page text. A web search surfaced a legacy CommonLook page titled Replace Text, but it returns a 301 to allyant.com/commonlook-pdf-replace/ which returns 404 page not found, so it is a dead legacy page and not current documentation. Release notes give current version CommonLook PDF 5.21, release date August 12, 2026, with 5.20 on May 4, 2026 and 5.19 on February 10, 2026. The July 8, 2025 company news article confirms the AI-driven web-based edition and the simplified and advanced editors, which supports the summary date. The privacy policy still says the services are hosted in the United States and/or Canada, and the product page still offers only a seven day trial request and a demo request, so watchOutFor and the quote only pricing stand. The edit job stays unresolved: no first party source documents an operation that changes existing page text, images or layout, and the tag, structure and metadata editing the vendor documents is the already scoped accessibility job, with the Remediator Tip as the vendor own statement that physical appearance does not change. Scoping edit to tag editing would duplicate the accessibility job. The job is not removed because the vendor does not say content editing is unavailable. A first party page documenting an operation that changes visible text, images or layout, or a vendor statement that this is not supported, would settle it. The summary, tagline, watchOutFor, platforms, interfaces, editions, pricing and accessibility scope were checked and left unchanged.

  273. 7a053ac

    The Documenso record, in the edit job, the owner field and the licence sentence

    What it said
    The edit job was unresolved on two earlier blockers that rested on the user guide pages of the time. The owner field was empty even though the vendor legal entity is named in its own terms and privacy policy. The summary also said the Business and Enterprise editions are paid licences, which the current licensing pages do not support: they describe a Community Edition under AGPL-3.0 and a single commercial Enterprise Edition, and the phrase business edition appears only inside the Fair Use Policy as a description of enterprise-level support.
    What it says now
    The edit job stays unresolved, now re-audited against the full current documentation corpus at https://docs.documenso.com/llms-full.txt, HTTP 200 with 1,031,997 characters over 29,883 lines, which contains no occurrence of edit the PDF, edit PDF, replace text, modify text, text editor, PDF editor, content editing or whiteout, no occurrence of merge at all, compress only in a database backup command, redact only in support instructions to redact sensitive information from logs, and add text only for Text fields that signers fill in. The upload page says PDFs are used as-is and DOCX is converted to PDF on the server during upload, that the converted PDF is what recipients sign, that the maximum is 50MB and configurable for self-hosted, and that encrypted and password-protected PDFs and legacy .doc, ODT, RTF and image formats are not supported. The lifecycle page says a draft allows adding and removing recipients, assigning roles, adding, moving and configuring fields, setting signing order and configuring settings, which is workflow preparation. Templates store the PDF, recipient roles, field placements and settings and are created by uploading PDF files. So Documenso prepares and signs an uploaded PDF and does not edit its visible text, images or layout, and document preparation and template editing are configuration rather than content editing. The owner field is now Documenso, Inc., the entity named at https://documenso.com/terms and https://documenso.com/privacy. The summary now says the self-hosted Community Edition is free under AGPL-3.0 while the Enterprise Edition is the paid commercial licence for proprietary use and support, with the hosted tiers Free at five documents a month, Individual at 25 USD per month, Teams at 40 USD per month including five users with additional users at 8 USD per month, and Platform at 250 USD per month with Enterprise add-ons, which the current pricing page still lists. The contribution pause in watchOutFor is confirmed by README and CONTRIBUTING, which say the project no longer accepts external pull requests aside from trusted contributors and directs contributions to detailed issues. Repository metadata confirms AGPL-3.0, default branch main and latest release v2.20.0 published 2026-10-08. The EU data residency claim is confirmed by the corpus, which says cloud hosting is in EU data centers and that self-hosted deployments choose their own storage. Platforms, interfaces, editions, pricing model, priceFromGbp, the sign scope and the claims array were checked and left unchanged.

  274. 7a053ac

    The Documenso record, in the summary licence sentence and the internal notes

    What it said
    The previous entry on this record concluded that the Business Edition was not a product edition and that the phrase appeared only once, inside the Fair Use Policy, then changed the summary to name a single Enterprise Edition. That conclusion was wrong. It came from searching only the documentation corpus at https://docs.documenso.com/llms-full.txt and the docs licensing pages, and from treating those as the whole of current licensing evidence.
    What it says now
    The Business Edition is a current self-hosted licence. The pricing page at https://documenso.com/pricing, fetched on 2026-10-16 with 592,062 bytes of source and 5,703 characters of visible text, contains three FAQ questions using the term, What kind of support is included with Business Edition (BE), How do I migrate from Community Edition (CE) to Business Edition (BE), and What happens if my needs grow beyond Business Edition (BE) how easy is it to upgrade. Its FAQ answers are not in the served source, but the page links Explore Self-Hosting Licenses to https://documenso.com/self-hosted, and that page, fetched on 2026-10-16 with 601,796 bytes of source and 6,547 characters of visible text, presents a current table headed Choose your Edition with columns Community Edition (CE), Business Edition (BE) and Enterprise Edition (EE), rows for Pricing Free Forever, Fixed Yearly License and Tailored Pricing, Intended Use Testing Evaluation Internal POCs, Serious Production Use and Large Scale Mission Critical, and Application Support Community Only, Reliable Support and Priority Enterprise Support, with SLAs, Roadmap Priority, Tailored License Terms and AATL Compliant Certificate No for Community, No for Business and Yes for Enterprise, and says to choose the edition that matches the level of support, guarantees and commercial relationship you need, from Community to Business to Enterprise. The summary now says self-hosted licensing is sold in three editions, the free Community Edition under AGPL-3.0, the Business Edition on a fixed yearly licence for serious production use, and the Enterprise Edition with tailored pricing, SLAs and priority support, while hosted plans run from a free tier of five documents a month through Individual and Teams to Platform. The watchOutFor sentence about the AGPL licence and the paid Business and Enterprise editions mattering for commercial self-hosting was already correct and is unchanged, and the owner stays Documenso, Inc. The dated internal notes keep the earlier wrong claim and add this correction rather than rewriting it, and record that the docs licensing pages present only Community and Enterprise while the self-hosted licensing page and the pricing FAQ present Business Edition as current, which is a documentation gap or older naming rather than evidence that the edition is gone. The edit job remains unresolved because nothing read in either pass documents editing visible PDF page content, and the full docs corpus still contains no occurrence of edit the PDF, replace text, modify text, PDF editor, content editing or whiteout. Hosted and self-hosted pricing figures are unchanged and nothing is published, deployed or contacted.

  275. 7a053ac

    The ICEpdf record, in the sign job, the licence field, the tagline and summary, and watchOutFor

    What it said
    The sign job was unresolved on two blockers written when the vendor site was the only source and every repository fact had been withdrawn. The licence field was empty although the code is Apache-2.0, and watchOutFor said the repository facts could no longer be verified from an official source. The earlier search had missed the maintained project because it lives under the maintainer personal account rather than an ICEsoft organisation page.
    What it says now
    The sign job is scoped from the maintained repository at https://github.com/pcorless/icepdf, which identifies itself as ICEpdf, carries Apache-2.0 headers naming ICEsoft Technologies Canada Corp. and Patrick Corless, and whose latest release read on 2026-10-16 is v7.4.1 published 2026-05-10. The source tree contains an acroform signature package with SignatureDictionary, SignatureFieldDictionary, SignatureHandler and SignatureManager, the signer handlers Pkcs12SignerHandler for a PKCS#12 keystore and Pkcs11SignerHandler for PKCS#11 tokens, and a DocumentSigner that writes a signed digest and a PDF 32000-1 12.8.1 byte range into the SignatureDictionary /Contents and /ByteRange entries; Pkcs7Generator creates a PKCS#7 signature from an X509 certificate chain and private key through BouncyCastle CMS; the examples/signatures module documents Pkcs12SignatureCreation with a keystore file, certificate alias and password, a hasPermissionToSignDocument check, a SignatureWidgetAnnotation added to the interactive form, a BasicSignatureAppearanceCallback with SignatureAppearanceModelImpl, plus Pkcs11SignatureCreation and SignatureVerification; SignatureType distinguishes SIGNER and CERTIFIER; verification covers adbe.pkcs7.detached, adbe.pkcs7.sha1, ETSI.CAdES.detached and adbe.x509.rsa_sha1 subfilters with certificate, CRL and OCSP verification classes; and the release notes document v7.3.0 adding a digital signature annotation tool with certificate selection, appearance edits and signature image inclusion, v7.3.2 adding PKCS#11 signature selection, and v7.4.0 adding TSA-configured signing with the signed timestamp attached to the signing certificate plus time-stamp validation and verification improvements. Drawn or handwritten signatures are not claimed, only the signature image placed in the appearance, and signature widget annotations are distinct from the link annotations the edit job covers. The licence field is now Apache-2.0. watchOutFor now records the maintained repository and its releases while keeping the site staleness, and explains that the earlier withdrawal is reversed because the project is under the maintainer personal account which the earlier search did not find. The tagline and summary now name the documented signing capability, with PKCS#12 and PKCS#11 signer handlers, configurable appearance and optional TSA timestamps. Platforms, interfaces, the self_hosted edition, pricing, priceFromGbp, the claims array, dataResidency and the view, extract, convert and edit scopes were checked and left unchanged.

  276. 7a053ac

    The Microsoft Edge PDF record, in the primary view job, the name, the owner and licence fields, the summary, watchOutFor and the Windows and macOS editions

    What it said
    The primary view job carried no scope, so the documented viewing surface and its behaviours were unrecorded. The owner and licence fields were empty, openSource was unset, and the record name said PDF reader and editor although the record itself, correctly, says the product cannot edit. The record also described the reader as if it were a single engine, while Microsoft documentation read on 2026-10-16 shows the built-in reader transitioning to an Adobe Acrobat-powered engine, with the policy that selects the older reader deprecated from Edge 156 and planned to become obsolete from Edge 162.
    What it says now
    The primary view job is scoped from the Microsoft Learn PDF reader page, the NewPDFReaderEnabled policy page and the stable channel release notes archive, all read on 2026-10-16. The scope covers a built-in reader that opens local, online and embedded PDFs, annotates with ink and highlighting, works on Windows and macOS, has a pin-able toolbar with zoom, rotate, fit to page or width, jump to page and search, and documents table of contents, page view, caret mode browsing, inking and ink customisation, highlight, text notes, read aloud, basic form filling with JavaScript forms not supported, Purview protected files in the same tenant on Windows and macOS and across tenants, Information Rights Management protected files, keyboard accessibility, high contrast and screen reader support, and viewing and validating certificate-based digital signatures with Microsoft saying more signature scenarios are being worked on; the documented constraints are that XFA is a legacy format not supported natively and can only be opened in IE mode when an administrator allows it through ViewXFAPDFInIEModeAllowedOrigins or ViewXFAPDFInIEModeAllowedFileHash. The scope also records the engine transition: the Adobe Acrobat-powered reader became the default experience for enterprise users on Windows from Edge 141, the NewPDFReaderEnabled policy is supported on Windows and macOS 111 and later and not on Android or iOS, is deprecated from Edge 156 because the reader experience has transitioned to the newer version, and is planned to become obsolete from Edge 162, while the newer engine adds higher signature verification coverage, MIP sensitivity label viewing, Undo and Redo for annotations on Windows, faster Read Aloud, Copilot Summarize and Explain actions and XFA via the IE mode policies. The scope states that it is the built-in viewer rather than a separate application, that editing and creating PDFs are not claimed, and that the legacy feature set is not claimed for the new engine or the reverse. The owner is now Microsoft, the licence is Proprietary with openSource false, and the name is Microsoft Edge PDF reader matching the vendor page title. The summary, watchOutFor and the Windows and macOS edition processing notes carry the engine transition. Platforms, the purview edition, pricing, priceFromGbp and the edit and sign scopes were checked and left unchanged.

  277. 7a053ac

    The Microsoft Lens record, in the primary ocr job and the convert job, the summary, the owner and licence fields and the two editions

    What it said
    The primary ocr job and the convert job carried no scopes, the owner and licence fields were empty, and the summary contained two claims the sources do not support: that the Actions modes applied across the product when the iOS page has no Actions mode, and that OCR made PDFs and Word files searchable when no page uses that word and OCR is documented only through the optional Word export.
    What it says now
    The ocr job is scoped from the surviving first party support pages read on 2026-10-16: OCR is documented as the optional Word export, which creates a Word file in the user OneDrive account that extracts all legible text from the captured image; the Android page also documents an Actions mode with a Text sub-mode that extracts text after selecting the preferred text language, plus Table, Read using Immersive Reader, Contact and QR Code sub-modes, while the iOS page lists only Whiteboard, Document, Business Card and Photo and has no Actions mode; both pages say GCC accounts do not support Image to Text, Image to Table, Image to Word, Image to PPT, Business Card or Upload to OneNote; the pages do not state where recognition runs, so on-device versus cloud processing is not claimed; and no accuracy or supported-language count is stated beyond the business-card note that it works best with English, German, Spanish or Simplified Chinese. The convert job is scoped to the documented exports: captures saved in Whiteboard, Document or Photo mode go to the Gallery or Photo Library, a PDF, OneNote, OneDrive, Word or PowerPoint, and on iOS also Outlook or the Mail app; the Word export applies OCR; the iOS page says Word, PowerPoint and PDF saves are in the OneDrive Documents folder, that PDFs can also be saved locally, and that the Word and PowerPoint outputs are real Microsoft Office documents with Word recognising handwritten, typed and printed text and the layout; Business Card mode saves a vCard to OneNote; on Android a previously Lens-created PDF can be edited by adding images, editing existing images or reordering them; and the scope states that it is not a general PDF converter and does not convert arbitrary PDFs to Office formats. Both scopes carry the retirement dates: retired from iOS and Android on 9 January 2026, no longer supported after 9 February 2026, and new scans impossible after 9 March 2026, with the retirement page stating that the app would be removed from the Apple App Store and Google Play Store on 9 February 2026 and that the recommended replacement is the OneDrive app, which the page says does not save scans locally on the device. The summary now attributes Actions to Android, describes the OCR-through-Word behaviour instead of claiming searchable output, and records the retirement dates; the owner is Microsoft and the licence is Proprietary; the editions record that Word, PowerPoint and PDF saves land in OneDrive while PDFs can also be local and that the pages do not state where recognition runs. The claims array, platforms, pricing, hasFreeTier, discontinued flag and note, bestFor and watchOutFor were checked and left unchanged.

  278. 7a053ac

    The Pdftools PDF Viewer SDK record, in the primary view job, the edit job, the watermarkOnFree field, the watchOutFor free-tier sentence and the four editions

    What it said
    The primary view job and the edit job carried no scopes even though the record prose claimed viewing and editing, the watermarkOnFree field was null although the vendor licence table says unlicensed pages are watermarked, and watchOutFor left the free testing route and the requested trial as a reconciliation for the reader rather than resolving it from the licence documentation.
    What it says now
    The view job is scoped from the current product page and Version 5 developer documentation: a TypeScript and JavaScript library that renders PDFs directly on the client without a server-side PDF library, with high-fidelity rendering, page navigation, zoom and scroll, fit mode and page layout mode, full-text search with highlighted and navigable results, opening from a URL, file or memory, saving and downloading as PDF or FDF with or without changes, printing with configurable DPI, and metadata and page information; two npm packages, PDF Web Viewer for a ready-made toolbar and sidebar interface and PDF Web SDK for a headless document view, with plugins, CSS theming, events and runtime localisation; tested support for Chrome 63+, Edge 41+, Firefox 55+ and Safari 11.0.3+; support for PDF 1.3 through 1.7 and 2.0 with the listed PDF/A levels; TypeScript and JavaScript use including React and Angular; native Android and iOS deployment including offline and air-gapped use from the product page; and no-licence use as free, view-only and watermarked with a 100 concurrent user ceiling. The scope keeps the boundary that it is the SDK own client-side viewer rather than a browser built-in and claims no server-side rendering. The edit job is scoped to adding, editing and deleting annotations of the documented types, filling form fields including text boxes, checkboxes, radio buttons and combo and list boxes, creating redaction annotations by selecting text or drawing areas and applying them to download a permanently redacted document, and saving or downloading as PDF or FDF, with the document boundary that the editing is annotation, form filling and redaction rather than editing existing page text, images or layout, and no form field creation or document-level JavaScript editing is claimed. The documentation is on Version 5 with 5.19.0 dated 14 October 2026, and both npm packages are at 5.19.0 with registry metadata modified on 2026-10-14. watermarkOnFree is now true on the licence-table evidence. watchOutFor now states the reconciliation: no licence key is free but view-only and watermarked with a 100 concurrent user ceiling, while Trial, Professional, Business and Enterprise are contact-sales feature sets with no watermark and no published price. The editions now record the client-side processing, the tested browsers, the mobile and air-gapped product-page claims and the unlicensed watermark. Platforms, interfaces, edition ids, pricing model, hasFreeTier, dataResidency, the redaction verdict and the two claims were checked and left unchanged.

  279. 7a053ac

    The Scanbot SDK record, in the primary ocr job, the platforms list, two new editions, the summary and watchOutFor

    What it said
    The primary ocr job carried no scope although the record prose already claimed OCR, so the inventory carried it as a prose-only signal. The platforms list said Android, iOS and web while the vendor documentation lists Data Capture Modules version 10.0.0, including the OCR module, for Linux and Windows Server, and watchOutFor already referred to Linux and server editions that the platforms and editions fields did not contain. The summary also said everything runs on the device, which the customer-server SDKs contradict.
    What it says now
    The ocr job is scoped from the vendor documentation read on 2026-10-16. The OCR module pages for Android, iOS and web, and the Linux and Windows Server Data Capture Module introductions, document an OCR engine that turns written text into machine-readable data from still images and a live camera stream and is the backbone of the Data Capture Modules, returning plain text, bounding boxes of paragraphs, lines and words, and confidence values per box. Android exposes OcrEngineManager with the Scanbot engine and an optional Tesseract engine through EngineMode.TESSERACT and the optional io.scanbot:sdk-ocr-tesseract dependency, with .traineddata language files plus osd.traineddata and no language data bundled. iOS offers a legacy Tesseract-based engine for non-Latin languages such as Arabic, Japanese, Chinese, Russian, Greek and Korean and a machine-learning engine that is faster and more accurate but Latin-only and needs no training data, with an optional bundle containing English and German examples. The web module returns the same text, bounding-box and confidence results. The built-in PDF generator can be combined with the OCR module to create searchable PDF files from a Document object or from images using PdfConfiguration defaults. No accuracy figure or supported-language count is stated on those pages, so none is claimed, and no hosted OCR is claimed because the vendor documents the SDK as completely offline with no networking code. Platforms now list Android, iOS, web, Windows and Linux, and two customer_server editions record the Windows Server and Linux deployments with their documentation sources and read dates, matching the watchOutFor sentence that already mentioned Linux and server editions. watchOutFor now names Linux and Windows Server precisely. The summary now says processing runs in the integrator own app, browser, Windows Server or Linux deployment while the vendor documents the SDK as offline with no networking code, replacing the earlier device-only sentence. Pricing was re-read on the contact-sales page that the pricing URL redirects to and still says a fixed annual fee generally starting at 2,500 USD for simple use cases and that the product is not free for production use, so the recorded 1,854.25 GBP per year figure, the year basis and hasFreeTier false are unchanged. The acquisition banner still says Scanbot SDK has been acquired by Apryse, matching the owner. The convert and extract scopes, processesLocally, selfHostable, dataResidency and the claims array were checked and left unchanged.

  280. 7a053ac

    The Qoppa Java PDF SDK record, in the edit, merge, convert, protect, extract, sign and compress jobs, the platforms list and the self_hosted edition

    What it said
    All seven remaining jobs carried no scopes even though the record prose claimed the capabilities and the inventory carried six of them as prose-only signals. The platforms list also said only self_hosted while the vendor solution and library pages state Windows, Mac, Linux and Unix for the Java libraries, and the single self_hosted edition said processing was unknown although the libraries are documented as running inside the integrator own application.
    What it says now
    All seven jobs are scoped from the vendor per-job solution pages and library pages read on 2026-10-16, while the overview still carries the notice that new licences are no longer sold and the functionality moved to the Apryse Java PDF, so the discontinued status, owner and licensing wording are unchanged. edit is scoped to jPDFEditor content editing, copy, delete, edit and move text, copy, delete, move and resize images, copy, delete, edit, move and resize shapes and paths, single or multiple objects, plus permanent redaction through jPDFEditor and jPDFProcess, annotations, form filling, digital signing and text search, with optional OCR and comparison modules named as separate. merge is scoped to jPDFAssemble assembling and disassembling, splitting, merging and combining with caller rules and bookmark and outline manipulation. convert is scoped to jPDFImages export of pages as JPEG, TIFF or PNG and import of images into new or existing PDFs with PDF 2.0 support, and to jPDFWeb PDF to HTML5 and SVG conversion, Word and Excel to HTML5 when combined with jOfficeConvert, with searchable text, fonts, image formats, vector graphics, annotations, links and form data, signature validation status and thumbnails, and no conversion-fidelity guarantee. protect is scoped to jPDFSecure encryption, permissions and passwords plus permanent redaction through jPDFEditor and jPDFProcess, with signing kept as the separate sign job and no algorithm, key-length or certification-level claim. extract is scoped to jPDFFields form fields and FDF, XFDF, XDP and XML import and export plus flattening, and to jPDFText text extraction for archiving, searching and indexing, with no OCR claimed there. sign is scoped to jPDFSecure creating and applying digital signatures and to jPDFNotes displaying and signing, with signature-field creation and self-signed certificate creation left to the desktop and server products the page names separately, and no algorithm, authority or timestamping claim. compress is scoped to jPDFOptimizer changing image compression quality, removing unused objects, merging duplicate objects and linearizing, with no percentage, quality-threshold or lossless guarantee. Platforms now list self_hosted, windows, mac and linux on the documented platform list, and the self_hosted edition records in_process processing with that document basis. The vendor own jPDFWeb statement that it was tested on JDK 1.6.37 and above is recorded as a dated vendor statement rather than a current compatibility claim, and Word and Excel conversion is claimed only to HTML5 when combined with jOfficeConvert because that product URL returned 404 when probed. The summary, tagline, watchOutFor, price fields, claims, discontinued flag and note, owner and licence wording were checked and left unchanged.

  281. 7a053ac

    The Qoppa Java PDF SDK record, in the discontinuedNote, the discontinuation claim and the internal notes, after a conflicting report about the live page

    What it said
    The record had relied on a September verification for the discontinuation notice. A reader reported on 2026-10-16 that the rendered vendor page no longer showed the notice and showed only Powered by Apryse, so the status needed to be reconciled against the live page instead of being carried from the earlier read.
    What it says now
    The live page was re-measured three ways on 2026-10-16 and still states the discontinuation. The served HTML, 64,467 bytes, contains the sentence We are no longer selling new licenses for these products. We have moved their functionalities and more to the Apryse Java PDF in a static div with id banner, and a headless Chrome 155 render at 15,000 ms produced 3,754 bytes of visible text containing the same sentence, with renders at 20,000 ms and 30,000 ms identical and still single-occurrence, so it is not a rotating slider. The same page lists all the libraries in the present tense and the footer reads Powered by Apryse. The notice appears on the overview and on the jPDFAssemble, jPDFOptimizer, jPDFSecure and jPDFText pages but not on the jPDFEditor page or the company About Us page, which may explain a check that missed it. The decision is option (c): a current first-party source still states the discontinuation, so the discontinued flag, the no-longer-sold licence wording and the null price are retained, while the provenance is corrected and the present-tense listing is recorded. The discontinuedNote now carries the live banner, the date, the fact that libraries are still listed, and the trigger that if the banner disappears on a future read, the flag, note, tagline, summary, watchOutFor, licence string and null price must be revisited together. The claim now records the 2026-10-16 re-check, the served-or-rendered method, the static banner element and the conflicting report rather than silently overriding it. The internal notes add a reconciliation entry with the exact measurements and the evidence hierarchy: current live first-party page, served and rendered, outranks the earlier snapshot, and one non-reproduced observation does not overturn three reproductions. Owner, licence and pricing were rechecked: Powered by Apryse supports the Apryse owner, the licence string still quotes the notice, no price or buy control appears on the overview, jPDFSecure or jPDFEditor pages, so priceFromGbp stays null. The seven job scopes describe documented capabilities without asserting current availability and are unchanged, as are the summary, tagline, watchOutFor, discontinued flag, platforms and edition processing. No disposition changes were needed, so the unresolved and verified-current counters are unchanged.

  282. 7a053ac

    The Syncfusion PDF record, in all nine remaining jobs, the summary OCR sentence, the platforms list and the editions

    What it said
    All nine remaining jobs carried no scopes even though the record prose claimed every one of them and the inventory carried six as prose-only signals. The summary also attributed Tesseract-based OCR to the PDF Library when the vendor publishes the OCR Processor under Data Extraction as a separate library. The platforms list omitted web even though the product FAQ names Blazor, ASP.NET Core and ASP.NET MVC as supported web platforms, and all four editions said processing unknown although the library runs inside the customer application.
    What it says now
    Nine scopes are written from the current documentation read on 2026-10-16. edit covers document creation and modification: text, images, tables and shapes, open, modify and save existing PDFs, AcroForms and XFA creation, filling and flattening, bookmarks, annotations, attachments, layers, hyperlinks, watermarks, barcodes, linearized PDFs, and adding, removing and rearranging pages, stamps, image insert/delete/extract/replace and attachments from the Edit PDF page, with no in-place text reflow or user interface claimed. convert covers HTML, RTF, Word, Excel, PowerPoint, SVG and XPS to PDF with the stated Blazor server-side limitation for Excel and named unsupported RTF elements, plus images and TIFF to PDF, PDF to image and PDF/A, with no PDF to Office conversion or fidelity guarantee. merge covers multiple PDFs from disk and streams through PdfDocumentBase.Merge plus splitting and page import, with no claim about preserving signatures or form data. sign covers adding certificate-based digital signatures with PdfSignature using PFX, HSM, OCSP, CRL, Windows Certificate Store and ECDSA, plus validation, with no timestamping or long-term validation claim. protect covers encryption and decryption with advanced standards, permission settings and redaction that removes underlying text and graphics inside the rectangle during the Redact call, with the stated CJK-without-TrueType and complex-script limits and the Imaging package requirement on .NET Core. compress covers compressing an existing PDF, new document content and images with image quality, embedded font optimization, page content optimization and metadata removal, with no size reduction or lossless guarantee. extract covers text extraction per page or document with line, word and glyph details and layout-based ExtractText(bool), plus image extraction with ExtractImages and PdfDocumentExtractor, with no OCR claimed there. ocr covers the separate .NET OCR Processor library running Google Tesseract OCR on scanned PDFs and images with an inbuilt preprocessor for grayscale, deskew, denoise, contrast and binarization, with no accuracy figure, language count or hosted OCR claimed. accessibility covers tagged PDF creation with tags for text, images, shapes, form fields, annotations, tables and lists, PdfStructureElement and ActualText, recording the vendor statement that tagged documents are compliant with Section 508 (PDF/UA) and WCAG 2.0 (ISO 14289-1:2014) as a vendor claim rather than a certification, and claiming no remediation of arbitrary existing PDFs. The summary now attributes OCR to a separate OCR Processor library rather than to the PDF Library. Web is added to platforms and a web edition records the Blazor and ASP.NET Core and MVC targets named in the product FAQ, while the existing editions move from unknown to in_process because the library runs inside the customer application; mobile is named in the FAQ only through .NET MAUI and UWP without naming Android or iOS, so no mobile platform value is asserted. Pricing was re-read and still shows the Document SDK at 1,199 USD per developer per year, consistent with the recorded 889.30 GBP conversion, and the Community License page still states the less than 1 million USD revenue, 5 developer, 10 employee and 3 million USD outside-capital limits already recorded in watchOutFor. The tagline, watchOutFor, claims, owner, licence string, hasFreeTier, price fields and dataResidency note were checked and left unchanged.

  283. 7a053ac

    The Power PDF record, in all eight remaining jobs, the watchOutFor mobile sentence, the summary, the mobile edition and the internal notes

    What it said
    All eight jobs carried no scopes even though the record prose claimed every capability and the inventory carried six of them as prose-only signals. The watchOutFor sentence also said web availability on mobile was not evidence of native iOS or Android apps, while the current product page documents mobile support for iOS and Android users to read, annotate, markup, sign and share PDFs and the Standard page says Power PDF is available on iOS and Android. The record carried no release version.
    What it says now
    All eight jobs are scoped from the current first-party pages read on 2026-10-16. view covers opening and reading PDFs on the documented desktop, web and mobile surfaces, including opening PDFs from Chrome or Edge into Power PDF and reading on the iOS and Android apps. edit covers creating, editing and compiling PDFs, text cut/copy/paste and right-to-left editing, moving pictures, annotations and stamps, page management and document assembly, comments and markup, form detection and form field creation and editing, with mobile documented as read, annotate, markup, sign and share rather than desktop-equivalent editing. convert covers PDF to and from Word, Excel, PowerPoint, JPG, HTML and more, PDF to Word Excel PowerPoint XPS and XPS to PDF, conversion without Microsoft Office, OCR-assisted scan conversion, and PDF/A and PDF/UA creation, with no fidelity guarantee. merge covers merging to a single PDF from multiple files, splitting by page range, content, bookmarks and size, and batch creation where the editions include them, with no claim about preserving signatures or forms. protect covers encrypted passwords and permissions with AES 128 and 256, redaction that removes text and graphics with advanced redaction properties, Inspect Document metadata removal, IRM and AD RMS, the EU Trusted List and SharePoint protection, and keeps the verified tier boundary that redaction is an Advanced feature and is not listed in Standard. sign covers stamp-based signatures in Standard, self-signed digital signatures to create and apply, signature validation management to view verify and navigate, DocuSign integration and embedded SignDoc workflows, distinguishing these from the separately sold SignDoc service and making no certificate-authority, timestamping or long-term validation claim. ocr covers the Tungsten OCR Engine v22, automatic detection of scanned PDFs, OCR proofreading and one-click capture with text detection, with no accuracy figure or language count. compress covers MRC compression for high-compression image PDFs and batch file size reduction, with no percentage, threshold or lossless guarantee. The watchOutFor mobile sentence is corrected to say that the iOS and Android apps are documented for reading, annotating, markup, signing and sharing rather than desktop-equivalent editing, a mobile edition records that documented surface, and the summary mentions the mobile app. The release is documented as 2025.3 in the release highlights and the 2025.3 documentation portal. The current pricing of 95 GBP Standard, 131 GBP Advanced and Business by quote with a five-user minimum was re-read and the recorded 95 GBP one-off price and headline are unchanged; the redaction tier claim was re-verified against the Standard and Advanced pages and is supported. The owner Tungsten Automation, the proprietary licence string and the platforms were checked and left unchanged.

  284. 7a053ac

    The Nutrient record, in the edit, protect, sign, compress, ocr and convert jobs, the owner field, the licence string, the platforms list, the editions and the watchOutFor affiliate sentence

    What it said
    All six remaining jobs carried no scopes. The owner field held only Orpalis Imaging SAS, one of four affiliates trading as Nutrient, while the watchOutFor sentence treated the operator as unresolved. The licence string named a Master Subscription Agreement that the current terms page did not surface in this read. The platforms list omitted windows and linux even though the edition records already carried them and the MAUI, server SDK and Docker and Kubernetes documentation supports them, and there was no self-hosted edition although the deployment page documents a self-hosted Document Engine and an offline-licence requirement for air-gapped use.
    What it says now
    All six jobs are scoped from the current solution pages read on 2026-10-16. edit covers text editing of existing text with font styles, sizes and colours, image manipulation, page management, annotations and the Content and Document Editor components, on the platforms the page lists, with no DOCX authoring because that is the separate Document Authoring product. protect covers pixel-perfect redaction with audit-ready tracking, pretrained and AI-assisted PII detection, OCR-backed redaction, custom rules and human-in-the-loop review, natural-language redaction through the AI Assistant, pattern and regex redaction through the REST Redaction API, and the Security page permission controls, password and permission assignment, watermarking and audit logging, with signing kept as the separate job and no encryption algorithm claimed because none is stated. sign covers electronic signatures in which users draw, type or upload a signature, X.509 certificate-based digital signatures, multiple signers, validation and integrity checks, timestamps, revocation data and long-term validation, metadata and audit trails, and the named PAdES, CAdES, eIDAS and ETSI standards, with no CA, TSA or certification-body claim. compress covers the Document Engine optimization endpoint, the .NET GdPicturePDFReducer API with more than 25 parameters, mobile on-device save strategies that remove unreferenced objects and rewrite the document structure, MRC compression, linearization and batch processing, on the platforms the page names, with no percentage, threshold or lossless guarantee. ocr covers turning image PDFs into searchable and selectable content, language and page selection, searchable PDF generation, text layer embedding with original layout preservation, mobile offline OCR, language detection and fallback, and more than 30 built-in languages, with Data Extraction kept as the separate component and the language count recorded as a vendor claim. convert covers DOCX, XLSX and PPTX to PDF with layout and style preservation, PDF/A and PDF/UA conversion, more than 100 file types including CAD through the .NET SDK, image conversion with optional OCR, resolution and compression control and metadata injection, with those counts recorded as vendor claims and no fidelity guarantee. The owner is now Nutrient, with the Impressum affiliates recorded in watchOutFor and the note that the contracting entity depends on the region. The licence string now states the component-based annual subscription model the current pricing page documents and no longer names the Master Subscription Agreement, which the terms page did not surface in this read. Platforms gain windows and linux, and a self_hosted customer_server edition records the self-hosted Document Engine and server SDKs with the offline-licence requirement for air-gapped deployment, while the local and remote editions record the WebAssembly zero-backend and hosted or managed deployment models. The summary, tagline, bestFor, claims, price fields and hasFreeTier were checked and left unchanged.

  285. 7a053ac

    The PrizmDoc record, in the view, edit, extract, convert, sign and ocr jobs, the platforms list and the two editions

    What it said
    All six remaining jobs carried no scopes. The platforms list omitted windows and linux although the documented server runs on Windows Server and Linux, and the editions carried platforms that do not match the documentation: the local edition listed mac, which the v14.10 system requirements do not support for the server, and described in-process processing although PrizmDoc Server is customer-deployed, while the remote edition listed desktop operating systems instead of web and api. The record also carried no release version or licence-change note.
    What it says now
    All six jobs are scoped from the current first-party pages read on 2026-10-16, with the documentation versioned PrizmDoc v14.10 Release. view covers the HTML5 viewer and its documented page, tile, thumbnail and text endpoints, the view, search, annotate, redact, print and download operations, the Office rendering option, large-document support and pre-converted viewing packages, and the tested client browsers, as a browser viewer with no fidelity guarantee. edit covers the 50+ file-type annotation support with layer-tracked revisions, management, merge and toggle, removing or burning in annotations while preserving underlying content, the template-based forms creation with signatures, initials, text boxes, dates, checkboxes, roles, input masks and tab order, the form-field detector for AcroForms and scanned TIFF or bitmap documents including its add-on and cloud inclusion, and text, diagonal text and image watermarks, with no in-place text or layout editing and redaction kept to the redaction field. extract covers the Search Contexts API text extraction the viewing documentation directs programmatic extraction to, plus the server-side search path for large documents, with no OCR or structured extraction. convert covers the content conversion REST API, its exact destination formats and the docx and xlsx conditions and Office-rendering licence dependency, multiple inputs and page selection only for pdf and tiff, password and rotation options, merging, and non-searchable to searchable PDF conversion, with no fidelity guarantee. sign covers freehand and typed electronic signatures with saved signatures reused across documents and sessions, with no certificate-based, timestamping, long-term validation or agreement-workflow claim. ocr covers the advanced OCR API, its page and zone OCR, orientation handling, per-character location and confidence, segmentation, character sets, rejection and confidence controls, expected data type and language, alternate results and validation lists, with more than 60 languages and an Asian add-on recorded as vendor claims and no accuracy figure. Platforms now list api, self_hosted, web, windows and linux. The local edition is customer_server on windows, linux and self_hosted with the documented server operating systems and the Windows Server 2016 deprecation; the remote edition is web and api on vendor_cloud with the AWS us-east-1 and 24-hour work-file facts and the scope of that retention statement. The internal notes record v14.10 and its September 2026 date, the new licence required for v14.0, the Apryse Server SDK Office renderer on Windows, HWP and HWPX conversion on Docker, and the dev.apryse.com legacy download path as operator evidence. The summary, tagline, bestFor, watchOutFor, claims, price fields, licence string, hasFreeTier and redaction field were checked and left unchanged.

  286. 7a053ac

    The Sejda record, in the edit, merge, compress, convert, sign, ocr and protect jobs, the owner field, the licence string, the watchOutFor free-tier sentence and the four editions

    What it said
    All seven jobs carried no scopes. The owner field was empty. The licence field was only Proprietary (EULA) with no entity or plan detail. The watchOutFor sentence said the free tier allowed three tasks a day for the whole product, while the web tools enforce three tasks per hour and the desktop enforces three tasks per day, and it omitted the conversion and single-file limits. The desktop page position, including its no longer support Free Download notice and the absence of perpetual licences, was not recorded.
    What it says now
    All seven jobs are scoped from the current first-party pages read on 2026-10-16. edit covers changing existing PDF text, adding text, images and links, creating and editing hyperlinks, annotating, filling and creating forms, with whiteout recorded as covering rather than removing content and scanned-page text left to the OCR job. merge covers combining PDFs and images, alternating and mixing pages, organising and reordering pages, and splitting by page range, bookmarks, two-page layouts, target file size and text occurrences, with the free combine limit of 30 files and 50 pages and no preservation claim. compress covers image optimization, unused font and image removal, JPEG quality 65 percent default with Good and Best options, 72 PPI default with 144, 288 and 720 PPI options and grayscale conversion, with no percentage or lossless guarantee. convert covers PDF to Excel and CSV, JPG, PNG and TIFF, PowerPoint, Text and Word DOC with Keep layout and Optimize for legibility, plus HTML, JPG and Word DOCX to PDF, with a free 20 page conversion limit and no fidelity guarantee. sign covers typed, drawn and uploaded-image signatures with over 10 handwriting styles and form filling, with no certificate-based, timestamping, long-term validation or audit-trail claim. ocr covers language selection for dictionary-based ambiguity resolution, searchable PDF and plain text outputs and an invisible text overlay at the correct locations, with free 10 pages and paid 100 pages and no accuracy or language-count guarantee. protect covers password protection with an admin password, Medium RC4 128 bit, Medium AES 128 bit and Strong AES 256 bit encryption and the listed reader permissions, with the vendor own warning that some readers ignore permissions. The owner is now Sejda BV, the Dutch company registered with the Chamber of Commerce Amsterdam under number 68286945 named in the terms effective 28 March 2026. The licence field records that entity, the terms effective date and the current plans: Web Week Pass at 5 GBP for 7 days as a one-time payment, Web Monthly at 7.50 GBP, Desktop+Web Annual at 63 GBP and Desktop Week Pass at 7.95 GBP, with the no-perpetual-licences position. The watchOutFor sentence is corrected to distinguish the web three tasks per hour from the desktop three tasks per day, add the one file per task, 20 page conversion and other tool limits, and record the desktop page no longer support Free Download notice against its legacy free-daily-limits copy. The web edition records encrypted upload, permanent removal after processing and two-hour deletion. The desktop editions record the current published platforms (Mac Intel and Silicon, Windows 64-bit and portable, Linux deb and rpm, Chromebook guide), the free-download contradiction and the paid desktop plans, while keeping the 2016 offline statement as a dated vendor statement. The summary, tagline, bestFor, claims array, priceFromGbp, priceBasis, hasFreeTier and platforms were checked and left unchanged.

  287. 7a053ac

    The Smallpdf record, in the compress, merge, edit, convert, sign, ocr and protect jobs, the owner field, the licence string, the pricing fields and summary, the watchOutFor and the editions

    What it said
    All seven jobs carried no scopes. The owner and licence fields were empty although the terms name Smallpdf AG in Z�rich. The record priced Pro at 9 GBP per month, while the current default pricing page displays 6.75 GBP per month billed as 81.00 GBP per year and the monthly toggle value could not be read, so the 9 figure was not verifiable. The platform, Pro-tier, eSign and encryption boundaries were unrecorded: macOS desktop support ended in 2021, direct text editing and scanned-PDF OCR are Pro features, signing is electronic only with Sign.com as a separate platform, and protection is AES 128-bit only.
    What it says now
    All seven jobs are scoped from the current first-party pages read on 2026-10-16. compress covers the free browser compressor and the Pro Strong Compression tier, with the without-losing-quality wording recorded as marketing rather than a lossless guarantee. merge covers combining, previewing, reordering and deleting pages with no preservation claim. edit covers adding text, shapes, images, highlights, sticky notes and drawings plus Organize mode, with direct text editing as a Pro feature and no OCR or layout-reflow claim. convert covers PDF to Word DOCX for free, other Office formats and the OCR option for scanned PDFs as Pro, with no fidelity guarantee. sign covers typed, drawn and uploaded-image electronic signatures with initials, dates, text and checkboxes on PDF, DOC, XLS, PPT, PNG and JPG files, distinguishing electronic from certificate-based digital signatures and naming Sign.com as the separate dedicated e-signing platform. ocr covers making scanned PDFs searchable and selectable, searchable PDF or searchable Word output and one-hour deletion, with no language list, language count or accuracy guarantee. protect covers AES 128-bit password encryption, TLS transfers, no account requirement and one-hour deletion with the password not saved, and makes no permission, AES 256 or enforcement claim. The owner is Smallpdf AG, Steinstrasse 21, 8003 Z�rich, Switzerland, and the licence field records the terms dated Z�rich, January 2026, governed by Swiss law, the Free, Pro, Team 2-19 seats and Business plans, the 7-day trial and the Sign.com Premium inclusion. The pricing fields now record 6.75 GBP per month with the month basis, verified from the default yearly pricing display as billed 81.00 GBP per year, and the internal notes explain that the Monthly toggle could not be activated so no monthly-billing figure is asserted; the summary replaces the stale 9 GBP sentence accordingly. watchOutFor adds the free-plan detail, the one-hour deletion, the Pro-only direct text editing and scanned-PDF OCR, the macOS desktop withdrawal after 1 October 2021 with Windows-only desktop coverage, the mobile scanner app redirect and its limited free access, and the Sign.com separation. The editions record the web upload with TLS and one-hour deletion and the default sharing-link caveat, the Windows offline statement, and the mobile app unknown processing. The claims array and hasFreeTier were checked and left unchanged.

  288. 7a053ac

    The Wondershare PDFelement record, in the edit, convert, merge, protect, compress, sign and ocr jobs, the owner and licence fields, the summary, the watchOutFor, the redaction evidence and the five editions

    What it said
    All seven jobs carried no scopes. The owner was the brand name rather than the legal entity, and the licence was only Proprietary with no plans, entity or trial detail. The recorded pricing URL returns 404 and the summary claimed OCR in 26 languages while the current OCR page says more than 20. The trial-limitation paragraph relied on archived pages and on a three-versus-five conversion page conflict that the current store page no longer supports. The five editions had no processing notes.
    What it says now
    All seven jobs are scoped from the current first-party pages read on 2026-10-16. edit covers text formatting, image move rotate flip crop and opacity, watermarks, hyperlinks, numbering, backgrounds, layers and multimedia, with scanned text left to OCR and the unregistered trial limited to Create, View, Protect and Organize Pages. convert covers PDF to Word, Excel, PowerPoint, Text, HWP and HWPX, images, PDF/A, HTML and EPUB and conversion to PDF from the listed formats, with no fidelity guarantee. merge covers merging PDFs and images, organizing the result by reorder delete and rotate, then compression, plus splitting into individual pages, with no preservation claim. protect covers AES-128, AES-256 and 128-bit RC4, open and permission passwords, the permission list for low-quality printing, signing or commenting, and copying, plus marking then permanently removing and blacking out redacted content, with signing kept to the separate job and no enforcement claim beyond the page. compress covers high, medium and low resolution reduction, manual DPI and batch compression, with the without-losing-quality wording recorded as marketing. sign covers typed, handwritten and image electronic signatures with auto-sync, certificate-based digital signatures with batch signing on Windows and Mac, and cloud-only send-and-track eSign, with no certificate authority, timestamping or independent legal certification claim. ocr covers editable and searchable output, layout preservation, more than 20 languages, and page range and language selection, with the language statement recorded as the vendor claim and no accuracy figure. The owner is now Wondershare Technology Group Co., Limited, with WONDERSHARE GLOBAL LIMITED in Hong Kong and WONSON GLOBAL ENTERPRISES INC in Canada recorded as operating subsidiaries, and the licence field records the current Yearly US.99, 2-Year US.99 and Perpetual US.99 plans, the 30-day money-back guarantee and the watermarked unregistered and registered 14-day trials. The summary adds the PDFelement 13 release and corrects the OCR claim. The watchOutFor now records the current per-feature trial limits from the store comparison, including eSign 1 time, AI 10 uses and Combine or Batch up to 3 files within 20 MB, replaces the archived three-versus-five page conflict, and keeps the Smart Redact and residency cautions. The redaction evidence now cites the live Smart Redact page with the permanent-removal and black-out wording and keeps the caveat that the wording uses hide once. The editions record in-process desktop, vendor-cloud web with cloud-only eSign sending, and unknown mobile processing. The claims array, hasFreeTier, watermarkOnFree, dataResidency, priceFromGbp and priceBasis were checked and left unchanged.

  289. 7a053ac

    The PDF2Go record, in the convert, compress, merge, protect, ocr, edit and sign jobs, the owner and licence fields, the platform set and editions, the summary and the watchOutFor

    What it said
    All seven jobs carried no scopes. The owner field was empty although the imprint and terms name QaamGo Web GmbH in Germany, and the licence was only Proprietary. The platforms included android and ios although the site publishes no native mobile apps. The record field priced the product at 6.85 GBP per month while the internal notes said no price could be captured, an unresolved contradiction in the record itself. The watchOutFor repeated an older 200-versus-400-file and 64-versus-8-GB conflict from help pages that now agree on 64 GB. The desktop page route used by the record was dead at /desktop, the OCR route at /ocr-pdf was dead, and the summary said well over a hundred tools where the site says 50 apps.
    What it says now
    All seven jobs are scoped from the current first-party pages read on 2026-10-16. convert covers PDF to editable Word, images, Excel, PowerPoint, Text, PDF/A and EPUB and to PDF from Word, images, Excel, EPUB, PowerPoint, DJVU and web pages, with the PDF to Word page own warning that designed layouts and scans may move and no fidelity guarantee. compress covers Basic and Strong compression, six dpi presets from 20 to 300, target-size mode with a 10 MB email preset, greyscale and image removal, with no lossless or size-reduction guarantee. merge covers combining two or more files in the chosen order, output page size and orientation, naming the output, and the separate split and organise apps, with no preservation claim. protect covers separate open and permission passwords with protection embedded in the file, with no encryption algorithm, permission list or reader-enforcement claim. ocr covers adding a searchable text layer to a scanned PDF and selecting one or more document languages, with no language count or accuracy claim. sign covers drawing, uploading or typing a signature, placing, resizing and rotating it, up to 50 MB without an account, and records the page explicit statement that this is not a certificate-based digital signature. edit covers adding text, images and shapes and filling forms including look-like forms and writing over scans, with changing existing body text left to the Word conversion route and no OCR or layout-reflow claim. The owner is QaamGo Web GmbH, Fritz-Reichle-Ring 2, 78315 Radolfzell, Germany, HRB 711602, CEO Jens-Peter Bierkandt, and the licence field records the German-law terms, VAT-inclusive prices, 12-month pay-as-you-go credit validity, expiring subscription credits, the 14-day money-back guarantee and the Starter, Pro and Business plans. The platforms are now web, windows, mac and linux, with android and ios removed because no native mobile apps are published; the web apps are recorded as browser use on phones. The pricing field remains 6.85 GBP per month, using the previously recorded conversion of the current EUR 8 Starter price rather than a new rate, and the internal notes now record the current EUR 8, 28 and 85 plan figures, the Save 19 percent yearly toggle whose amounts were not rendered, and the desktop app included from Pro upward. The summary says 50 apps and the desktop app from Pro upward. The watchOutFor records the free tier as limited, ad-supported and without batch or AI, the 50 MB unregistered edit and sign limit against the 64 GB paid maximum, the Pro-upward desktop tier, the 24-hour access block and latest 72-hour physical deletion with no backups, and the credit distinctions. The dead /desktop and /ocr-pdf routes are replaced by the working /download/pdf2go-desktop and /create-searchable-pdf routes. The claims array, hasFreeTier, watermarkOnFree, dataResidency, priceFromGbp and priceBasis were checked and left unchanged.

  290. 7a053ac

    The Stirling PDF record, in the edit, merge, compress, convert, sign, ocr and protect jobs, the owner and licence fields, the summary, the watchOutFor and the six editions

    What it said
    All seven jobs carried no scopes. The owner was empty although the footer and terms name Stirling PDF, Inc. The licence string said only that the core is MIT with some separately licensed directories, without naming them or the hosted terms. The internal notes said priceFromGbp was null while the record field held 73.43 GBP per month, an unresolved contradiction. The summary said the self-hosted edition was free for up to five users, which the current pricing page does not support. The watchOutFor grouped OCR, Office conversion, compression and repair together as needing a server, while the documentation names a precise list. The editions had no processing notes and no record of the desktop, cloud and self-hosted modes.
    What it says now
    All seven jobs are scoped from the current first-party site and documentation read on 2026-10-16. edit covers the PDF Text Editor for existing text and images with its alpha warning that fonts and layout can shift, image add and extract, text and stamps, annotations, removal and metadata, with scans left to OCR. convert covers the full to-PDF and from-PDF matrix including Office, images, text, HTML, Markdown, email, eBooks, comics, PDF/A, PDF/X and PDF/UA, with image DPI 72 to 600 and no fidelity guarantee. merge covers ordering, table of contents, signature removal, the split methods, reorganise and rotate, and records that merging breaks digital signatures. compress covers the quality slider 1 to 9 default 5, target file size, grayscale, linearization and line-art options with no lossless guarantee. protect covers user and owner passwords, permission changes, sanitisation, flattening and form unlocking, with no reader-enforcement or algorithm claim and redaction left to the redaction field. sign covers drawn, typed, uploaded and phone-scanned visible signatures with the saved-signature limits and warns that the visible signature can be copied, and separately covers certificate signing with PKCS12/PFX, PEM or JKS, trust and revocation checks, RFC 3161 timestamps and validation, with no certificate-authority or timestamp-authority guarantee. ocr covers required language selection, auto, force and strict modes, processing options and the Tesseract language-pack position, with no accuracy or universal language-count claim. The owner is Stirling PDF, Inc., and the licence field records the MIT core with the separately licensed app/proprietary, app/saas, engine and frontend proprietary directories and the hosted Terms of Service v1.1.0 dated 8 August 2026 under Delaware law. The pricing fields keep 73.43 GBP per month as the previously recorded conversion of the current USD 99 Team plan, with the Free, Team, Enterprise and Processor plans recorded in the licence field and the internal notes, and no new rate invented. The summary says 50+ tools and replaces the five-user sentence with the current Free, Team, Enterprise and Processor plans. The watchOutFor records the exact cloud-or-server tool list, the Local Only cloud badge, the OCR language-pack position with the standard and ultra-lite image differences, the certificate and timestamp configuration dependency and the verified Java 25 requirement for the 2.x server jar. The editions record the desktop local-engine modes, the hosted Stirling Cloud, the private API on the self-hosted server and the customer-environment position, with processing left unknown because the record has a measured self-hosted locality probe and the page publishes that verdict rather than a Mixed badge. The claims array, hasFreeTier, redaction and dataResidency were checked and left unchanged.

  291. 7a053ac

    The PDF Architect record, in the edit, convert, protect, view, merge, sign and ocr jobs, the owner and licence fields, the price and summary, the watchOutFor and the two editions

    What it said
    All seven jobs carried no scopes. The owner was the Avanquest brand rather than the operating entity named in the imprint. The licence was only proprietary with no entity, plans or dependencies. The price field held GBP 59 per year while the current pricing page sells Pro at GBP 94 per user per year. The summary called the paid plans Professional and Business while current vendor surfaces use three different sets of names. The free-tier limits, version, Office dependency and activation behaviour were not recorded, and the OCR language count was not qualified.
    What it says now
    All seven jobs are scoped from the current first-party pages and help articles read on 2026-10-16. view covers the free PDF reader and navigation. edit covers direct text and content editing with font and style changes, with scans left to OCR and no layout guarantee. convert covers Create and Convert directions for Word, Excel, PowerPoint, images, HTML, RTF and PDF/A, with the Microsoft Office plug-in dependency and no fidelity guarantee. merge covers combining multiple files and attachments into one PDF and the page-organisation features, with no preservation claim. protect covers the password to open with a selectable encryption level, permission restrictions with a separate password, and manual and search redaction that permanently deletes information after an Apply Redaction step, with no specific algorithm claim and permanent deletion kept as the vendor claim. sign covers the Fill and Sign module for placing signatures and initials, the separate E-Sign service for preparing, sending and tracking signatures with a ceremony manifest, and digital signatures in the Secure module, with 50 e-signatures a year on Pro and unlimited on Business, and no independent legal-validity or certificate-authority claim. ocr covers Recognize Document and Auto and Manual OCR that make scanned text editable, lists the 24 languages in the help article while flagging the Business card 100+ claim as a conflicting source, and makes no accuracy guarantee. The owner is now Avanquest pdfforge GmbH, Axel-Springer-Platz 3, 20355 Hamburg, HRB 185235, European tax ID DE183155981, with Eric Gareau and Stephen Siegel as managing directors, and the licence field records the entity, the trademark position, the current Free, Pro at GBP 94 and Business at GBP 146 per user per year plans, the 30-day money-back guarantee, the three-way plan naming across the pricing page, product page and the older PDF Architect 9 help article, the Office 2007 or later requirement for plug-in conversion and the internet requirement for activation and validation. The price field is now GBP 94 per year, priced natively in GBP with no conversion, and the summary records the current plans and the naming conflict. The watchOutFor records the free-card versus comparison-table conflict, the plan-naming conflict, the Pro and Business e-signature allowances, the Office dependency, the offline reversion to PDF Architect Free, the OCR 24-versus-100+ conflict and the whiteout versus redaction distinction. The editions record in-process Windows desktop processing with activation checks and the separate vendor-cloud E-Sign service. The claims array, hasFreeTier, watermarkOnFree, redaction and dataResidency were checked and left unchanged.

  292. 7a053ac

    The Soda PDF record, in the edit, convert, merge, protect, compress, sign and ocr jobs, the owner and licence fields, the price fields, the version, the platform set and the editions

    What it said
    All seven jobs carried no scopes. The owner was the Avanquest brand rather than the controller named in the privacy policy. The licence was only Proprietary. The record held GBP 10 per month from an old checkout reading while the current plans page publishes no prices at all and says See prices, and the internal notes themselves said priceFromGbp was null. The version was unrecorded. The platforms listed web, windows, mac, linux, ios and android as if all were native, although the desktop app is Windows-only and the other platforms are browser access. The free limits, the whiteout wording, the E-Sign versus digital signature distinction and the current plan naming were stale or unrecorded.
    What it says now
    All seven jobs are scoped from the current first-party pages and help articles read on 2026-10-16. edit covers direct text editing and formatting with scans left to OCR and no layout guarantee. convert covers the documented PDF to and from Word, Excel, PowerPoint, image and HTML directions with the Office plug-in dependency and no fidelity guarantee. merge covers combining, reordering and saving merged PDFs with no preservation claim. protect covers password and permission controls, records the vendor statement that 256-bit AES is the strongest PDF encryption and that 128-bit AES is no longer supported by PDF 2.0, and keeps the Secure module Redaction as the vendor permanent-deletion path while leaving the separate whiteout tool out of redaction. compress covers Regular and High compression with the vendor estimated 40 and up to 85 percent reductions, with no lossless guarantee. sign covers the E-Sign module Sign your Document and Request Signature routes and the separate local certificate-based Digital Signature, with the article statement that E-Sign authenticates outside the local system and the local certificate can only be verified on the machine that created it, and makes no independent legal-validity, certificate-authority or timestamping guarantee. ocr covers the OCR module and documents the 25-language list, with no accuracy guarantee. The owner is Avanquest Software SAS, Immeuble Adamas, 2 rue Berthelot, 92400 Courbevoie, France (RCS Nanterre), and the licence field records the Canadian contracting entity 7270356 Canada Inc. (Avanquest Software), the subscription-only plans, the hidden pricing, the free limits, the desktop dependencies and the E-sign retention facts. The price fields are null with the basis unknown because the plans page publishes no amounts and the checkout rendered no readable price, so the previous GBP 10 monthly reading is recorded as unverified rather than asserted. The summary and watchOutFor record Soda PDF 15, the plan naming conflict with the older Soda PDF 14 Standard/Pro/Business article, the hidden pricing, the free limits, the online-only AI and cloud features, the AES statement, the whiteout versus redaction wording and the E-Sign versus digital signature separation. The platforms are now web and windows, and the editions are vendor-cloud web processing with 24-hour deletion and multi-region data and in-process Windows desktop processing with activation and Office dependencies. The redaction evidence cites the current help article again with the stock-copy caveat and the whiteout distinction. The claims array, hasFreeTier, watermarkOnFree, batchSupported and dataResidency were checked and left unchanged.

  293. 7a053ac

    The PDF Candy record, in the convert, edit, compress, merge, ocr, protect and sign jobs, the owner and licence fields, the price basis, the summary, the watchOutFor and the two editions

    What it said
    All seven jobs carried no scopes. The licence field was empty although the service is commercial, has a Windows desktop app and sells annual and lifetime plans, and the owner was the shorter brand name rather than the full company named in the privacy policy. The price field held GBP 4 per month without saying that this is the displayed monthly equivalent of the Desktop + Web Yearly plan, while Web Monthly is GBP 6 per month and Lifetime is GBP 99 one-off. The free-tier page, hourly, file-size, batch and desktop-operation limits were not recorded, the desktop was described as performing the same kinds of jobs without the 35-plus versus 98 tool gap, and the signing, OCR and whiteout boundaries were not stated.
    What it says now
    All seven jobs are scoped from the current first-party pages read on 2026-10-16. convert covers more than twenty to and from PDF directions across images, eBooks and documents, with no fidelity guarantee and no format beyond those listed. edit covers text adding, deleting, formatting and changing with standard-font recognition and the larger-edit layout-shift warning, image adding, highlighting, page reordering and insertion, annotations and signing, with scan text left to OCR. compress covers Maximum and Recommended modes with no lossless or percentage guarantee. merge covers joining several PDFs, ordering, images and page reordering without watermarks, with no preservation claim. ocr covers all-page recognition into editable text with the blurry-text caveat and no language list or accuracy figure. protect covers password encryption that mainly restricts opening, separate permission restrictions for printing, copying and editing, the seven-day share link and two-hour deletion, with no specific algorithm claim and whiteout kept out of redaction. sign covers text, freehand and image signatures with no certificate-based or legal-validity guarantee. The owner is Icecream Apps Limited, Cyprus, the controller named in the privacy policy effective 22 July 2025, and the licence field records the plan structure, Paddle, the refund position and the deletion facts. The price fields keep GBP 4 per month with the explicit note that it is the monthly equivalent of the Desktop + Web Yearly plan billed annually, against a listed GBP 18 per month, with Web Monthly at GBP 6 and Lifetime at GBP 99 one-off. The summary says more than ninety tools, Desktop 3.20 with 35+ tools and the free desktop operations limit. The watchOutFor records the free page, hourly, file-size, batch and desktop limits, the yearly-versus-monthly price distinction, the 35-plus versus 98 tool gap, the whiteout covering result, the electronic-only signing and the OCR limits, and cites the security page EU and USA servers rather than an unsupported regions caveat. The editions record vendor-cloud web processing with two-hour deletion and in-process offline Windows desktop processing. The redaction evidence remains the tested covering result, with the vendor whiteout wording and the EU and USA server context added. The claims array, hasFreeTier, watermarkOnFree, maxFileSizeMb, batchSupported and dataResidency were checked and left unchanged.

  294. 7a053ac

    The pdfRest record, in the edit, convert, protect, ocr, merge, compress, sign and extract jobs, the owner and licence fields, the summary, the watchOutFor, the dataResidency and the two editions

    What it said
    All eight jobs carried no scopes. The owner was the Datalogics brand rather than the legal entity in the terms. The licence field was a short commercial description without the entity, plans, deployment structure, retention or the EU backend fee. The summary described the API Lab as a browser route in a way that implied a website surface when the product is an API and self-hosted toolkit. The watchOutFor described the deployment matrix and price plans but not the per-plan retention, upload limits, EU backend fees or the billing of POST, DELETE and polling calls. dataResidency was unknown although the standard cloud is United States-hosted and the EU backend is an optional routing choice.
    What it says now
    All eight jobs are scoped from the current endpoint documentation read on 2026-10-16. edit covers Add to PDF adding text, images, lines, rectangles, tables and attachments without disrupting underlying content, and makes no existing-text editing or layout claim. convert covers PDF to Word, Convert to PDF, Convert to PDF/A and the listed conversion endpoints with no fidelity guarantee. protect covers Encrypt PDF with a password to open, Restrict PDF with a permissions password for printing, copying and editing, and the Pro Redact PDF endpoint removing sensitive information by text or regular expressions, with the removal kept as the vendor statement rather than a test and no specific algorithm or reader-enforcement claim. ocr covers the Pro OCR PDF endpoint making scanned documents searchable and extractable, with no language or accuracy claim. merge covers Merge PDFs and Split PDF with no preservation claim. compress covers Compress PDF with no lossless or size-reduction guarantee. sign covers the Pro Sign PDF endpoint applying cryptographic digital signatures with PFX or certificate credentials and custom visual appearances, with no certificate-authority, timestamping or long-term-validation claim. extract covers Extract Text with style and positional information and Extract Images without altering the original, with no accuracy or reconstruction guarantee. The owner is Datalogics, Inc., 1207 Delaware Ave, Suite 1810, Wilmington, DE 19806-4743, under Delaware law, and the licence field records the Cloud, AWS and Container commercial structures, the Starter, Premium, Pro, Enterprise and Enterprise+ plans, the upload and retention differences and the EU backend fee. The summary records the API Lab console and the AWS and Container alternatives. The watchOutFor separates the cloud, AWS and container modes, records the United States default and optional EU backend, the AWS Marketplace or BYOL billing, the on-premises license key with a 14-day trial, the per-call billing including POST, DELETE and polling, the per-plan retention and upload limits, the Starter watermark and Pro tools, the deployment-dependent residency and the endpoint-specific caveats. dataResidency is now multi, and the editions record vendor-cloud API processing with United States and EU options and customer-server processing for AWS and Container. The redaction evidence cites the current Redact PDF endpoint with the untested-removal caveat. The priceFromGbp remains 7.48 GBP per month as the previous conversion of the current USD 9.99 monthly Premium price, with no new rate invented. The claims array, hasFreeTier, watermarkOnFree, maxFileSizeMb and batchSupported were checked and left unchanged.

  295. 7a053ac

    The SnackPDF record, in the compress, merge, edit, convert, extract, protect and ocr jobs, the removed sign job, the licence and openSource fields, the summary, the watchOutFor and the web edition

    What it said
    Seven real jobs carried no scopes and a sign job was tagged although the current site offers no signing tool and the signing routes return 404. The licence field was empty although the project publishes its AGPL-compatible distribution, Ghostscript and encryption-module licences. The owner field was empty and remains so because no operating legal entity is published; the old watchOutFor also said non-interactive flat forms were not detected while the current fill-form page says flat PDFs use positioned static marks, and it did not record the AGPL licence position or the network-delivery boundary of the local-processing claim.
    What it says now
    Seven jobs are scoped from the current tool pages read on 2026-10-16. compress records the Light, Balanced and Smallest profiles at 220, 150 and 96 DPI with no lossless guarantee. merge records the two-to-fifty PDF order and the rewrite caveat with no preservation claim. edit records form filling of existing AcroForm fields plus positioned static marks on flat PDFs, with live values and font limits, and no general text, image or annotation editing. convert records PDF to JPG and PNG at 96, 150 and 300 DPI with no Office or PDF/A claim. extract records stored-text extraction and page extraction in entered order, with scans left to OCR. protect records PDF 2.0 AES-256 with open and owner passwords, advisory permissions and the encryption-is-not-redaction boundary. ocr records printed-English recognition returning a page-by-page text file rather than a searchable PDF, with no accuracy guarantee. The sign job is removed because the site lists sixteen tools with no signing tool and the signing routes are 404. The licence field records the AGPL-compatible distribution, application AGPL-3.0, Ghostscript 10.06.0 AGPL-3.0-or-later, the MIT encryption module, the absence of a commercial Ghostscript licence, and the free independent browser-local project status with no accounts, storage, server processing, API, AI or email collection; openSource is set true. The owner is left unset because no operating legal entity is published and the terms say operator details have not been added. The summary and watchOutFor record the free AGPL project with no published legal entity, the absence of signing, the form-filling-only edit boundary, the image-only conversion, the stored-text versus OCR distinction, the page-rasterising redaction mechanism and its text-layer cost, the advisory permissions, the unlock limit, the local-processing boundary including application delivery and hosting request information, and the licence position. The web edition records in-process browser-Worker processing with the tested no-document-transmission claim and offline caching. The claims array, disclosure, processesLocally, requiresAccount, watermarkOnFree, maxFileSizeMb, batchSupported and dataResidency were checked and left unchanged.

  296. 7a053ac

    The PyMuPDF record, in the protect, view, extract, merge, edit, ocr and compress jobs, the owner and licence fields, the platform set, the editions, the summary and the watchOutFor

    What it said
    Seven jobs carried no scopes. The owner was the brand short form rather than Artifex Software, Inc., and the licence string did not record the dual AGPL or commercial position, the exclusive commercial agent, the contact address or the separate PyMuPDF Pro and PyMuPDF4LLM products. The platform set included self_hosted as if the library were a hosted service, and the watchOutFor did not record the separate Tesseract installation, the absence of a hosted service and API, the non-viewer nature, the default reading-order caveat or the document-dependent compression.
    What it says now
    Seven jobs are scoped from the current documentation read on 2026-10-16. protect records the Document.save() encryption method codes, document permissions and owner and user passwords, with no reader-enforcement claim. view records Page.get_pixmap() page and clip rendering at adjustable resolution, with no interactive-viewer claim. extract records get_text() plain, blocks, words, dict and rawdict modes, Markdown, key-value and table extraction, embedded image extraction and the reading-order recipes, with the documentation caveat that default output is the text as coded and may not be in usual reading order, and no OCR claim. merge records Document.insert_pdf() and Document.insert_file() with no preservation claim. edit records annotations added and modified through Page and Annot, notes that annotation insertion is PDF-only, and records inserted text and images, form fields and metadata, with no reflow or rewriting of existing body text. ocr records the Tesseract-backed OCR path through Pixmap.pdfocr_save(), Pixmap.pdfocr_tobytes() and Page.get_textpage_ocr(), with the external Tesseract installation and language data required and no accuracy or bundled-language-count claim. compress records the Document.save() garbage levels 0 to 4, deflate including Flate and experimental Brotli, use_objstms and the related image and font compression options, with no lossless or guaranteed size-reduction claim. The owner is Artifex Software, Inc., and the licence records the dual AGPL or commercial licence, Artifex as the exclusive commercial licensing agent for MuPDF, the San Francisco address, the absence of a published commercial price and the separate PyMuPDF Pro and PyMuPDF4LLM products. The platforms are now windows, mac and linux, with the self_hosted platform and edition removed because PyMuPDF is a Python library rather than a hosted service; the editions record in-process library processing with no hosted endpoint. The summary records version 1.28.2. The watchOutFor records the AGPL-or-commercial blocker, the absent hosted service, cloud API and managed endpoint, the non-viewer nature, the multiprocessing recommendation, the annotate-then-apply redaction and its changelog history, the subset CLI with redaction Python-only, the separate Tesseract requirement, the default reading-order caveat and the document-dependent compression. The claims array, redaction, dataResidency, openSource, selfHostable and interfaces were checked and left unchanged.

  297. 7a053ac

    The Telerik DPL record, in the edit, convert, extract, accessibility, merge, sign and protect jobs, the licence field, the platform set, the editions, the summary and the watchOutFor

    What it said
    All seven jobs carried no scopes. The platform set included self_hosted as if the libraries were a hosted service, and the editions carried an unknown self_hosted edition. The licence field described the DevCraft bundles in general terms without the current tiers, without stating that the bundle price is not a standalone DPL price, and without the 30-day trial. The accessibility claim was a single Generate accessible PDFs bullet with no per-output-path limits, the digital signature scope did not record the X.509 and CA boundary, the password settings did not record the supported AES256, AES128 and RC4 types or the user-access permissions, and the merge and extraction capabilities were unrecorded.
    What it says now
    All seven jobs are scoped from the current documentation read on 2026-10-16. edit covers PDF, Word and spreadsheet creation and editing, including text, images, shapes, tables, lists, forms, annotations and bookmarks and Word fields and hyperlinks, with no UI claim. convert covers PDF import/export to PDF and plain text, Word DOCX, DOC, RTF, HTML, TXT and Markdown to PDF and between supported formats, spreadsheet XLSX, XLSM, XLS, CSV, TXT and JSON plus PDF export, and the PDF/A and PDF/UA compliance targets, with no fidelity guarantee. extract covers the documented extract-and-search PDF capability, plain-text export and form data extraction, with no OCR claim. accessibility covers documented generation of accessible PDF output targeting PDF/A-1a, PDF/A-2a, PDF/A-3a or PDF/UA-1 with tagging strategy and structure trees, records that auto-tagging defaults off, and records the conformance report limits that WordsProcessing-to-PDF is partially tagged, SpreadProcessing-to-PDF is untagged and automated validation does not establish semantic quality, explicitly not a certification. merge covers PDF merge, split and append and content positioning through PdfStreamWriter plus Word document merge, with no preservation claim. sign covers digital signatures and validation with X.509 certificates, chains, trusted roots and revocation checks, timestamp servers and external signing, with certificate issuance and CA management outside scope. protect covers password protection and encryption with AES256, AES128 and RC4, user and owner passwords and user-access permissions, with the PDF/A encryption exclusion noted. The licence field records the five-library commercial suite, the current DevCraft UI, Complete and Ultimate prices per developer per year, the absence of a published standalone DPL price, the inclusion in DevCraft and other Telerik products, the 30-day trial and the Telerik commercial licence requirement. The platform set is now windows, mac and linux only, the three OS editions record in-process library processing with no hosted endpoint, and dataResidency is self_hosted because processing happens inside the host application. The summary and watchOutFor record the bundle-versus-standalone distinction, the accessibility limits, the X.509 signing scope and the no-hosted-service shape. The claims array, priceFromGbp, priceBasis, hasFreeTier and interfaces were checked and left unchanged.

  298. 7a053ac

    The PDF24 record, in the merge, compress, edit, sign, ocr and protect jobs, the owner and licence fields, the summary, the watchOutFor and the two editions

    What it said
    The six jobs carried no scopes. The owner and licence fields were empty even though the legal notice names geek Software GmbH and the online tools and Creator are free. The summary and watchOutFor did not record the current Creator version 11.31.1, the German server location and one-hour deletion, the no-paid-tier position or the optional AI invoice-recognition exception, and the edit, sign, protect and OCR boundaries were not stated.
    What it says now
    The six jobs are scoped from the current tool pages read on 2026-10-16. merge covers combining multiple PDFs, images, Word and other supported files with drag ordering, page mode, bookmarks and blank pages, with the without-losing-quality wording treated as marketing. compress covers reducing image resolution and quality and removing unneeded data with adjustable DPI and quality, with no lossless or percentage guarantee. edit covers writing on pages, inserting images, rectangles, circles and arrows and freehand drawing, with no existing-text rewrite claim because the tool itself suggests converting to Word. sign covers electronically signing by drawing, uploading or photographing a signature, with no certificate-based, cryptographic, trust-validation or timestamping claim. ocr covers searchable text recognition and PDF/A output with deskewed and cleaned pages and a long language selector, with no accuracy figure or language-count guarantee. protect covers AES-256 password encryption and restrictions on printing, copying, commenting and changing, with no reader-enforcement claim and redaction kept separate. The owner is geek Software GmbH, Einstein Palais, Friedrichstrasse 171, 10117 Berlin, HRB 100865 B, VAT ID DE248847214, with Benjamin Bohlmann as managing director. The licence field records proprietary freeware, the free and ad-funded position, the legal notice statement that the free services remain free, the no-subscription and no-premium terms, and the exclusion of Creator and Fax from the online terms. The summary and watchOutFor record Creator 11.31.1 released on 8 October 2026, the Windows-only offline Creator, the German servers, SSL and one-hour deletion with no backups, the optional AI invoice-recognition exception, the absence of a paid tier, and the edit, sign, protect and OCR boundaries. The web edition records vendor-cloud processing with German servers and one-hour deletion, and the Windows edition records in-process Creator 11.31.1 offline processing. priceFromGbp and priceBasis remain null and unknown because no paid tier was found, dataResidency stays eu and platforms stay web and windows. The claims array, redaction, hasFreeTier and watermarkOnFree were checked and left unchanged.

  299. 7a053ac

    The PDFsam record, in the merge, compress, edit, sign, ocr and protect jobs, the owner and licence fields, the summary, the watchOutFor, the platform fields and the three editions

    What it said
    The six jobs carried no scopes. The owner field was empty although the About page and EULAs name Sober Lemur S.r.l. in Italy, and the licence field did not record the current Visual and Enhanced prices or the Italian EULA entity. The internal notes attributed PDFsam Enhanced to LULU Software in Canada, which does not appear on the current EULAs or about page. The Visual edition was listed as Windows-only although the current download page ships Windows, macOS and Linux, and the Basic edition listed self_hosted as a platform although it is a desktop and CLI application. The summary described a web or online surface that does not exist, and the edition versions, trial terms and free-tool boundaries were not recorded.
    What it says now
    The six jobs are scoped from the current pages read on 2026-10-16. merge covers PDFsam Basic combining multiple PDFs with ordering and output naming and the Visual free Merge, Alternate Mix, Split by bookmarks and Split by size tools, with no preservation claim. compress covers Visual compression with adjustable image quality and resolution, records that Basic performs only incidental optimisation rather than true compression, and makes no lossless or specific size-reduction guarantee. edit covers the Enhanced editor modifying and adding text, annotations, images and links, redaction, watermarks, headers, footers, comments and shapes, with no fidelity or layout guarantee and redaction kept to its field. convert remains as previously scoped. sign covers Enhanced Secure digital-certificate signing in the Pro and Pro+OCR tiers, with no certificate-authority, trust-validation, timestamping or independent legal-validity guarantee. ocr covers Enhanced Pro+OCR Advanced recognition of images and scans into editable PDFs, including batch recognition, with no accuracy or language-count claim. protect covers Visual Protect and Unprotect with passwords and permissions and Enhanced Secure 256-bit AES and permission restrictions, with no reader-enforcement claim. The owner is Sober Lemur S.r.l., Via Rosemberg 4, 42021 Bibbiano (Reggio Emilia), Italy, VAT IT02771170350, REA RE-312122. The licence field records Basic as AGPL-3.0, Visual and Enhanced as proprietary under Sober Lemur EULAs dated 1 September 2023 and governed by Italian law, the Visual subscription at GBP 29 per user per year with recurring billing and VAT included and the GBP 49 perpetual option, and the Enhanced Standard, Pro and Pro+OCR Advanced tiers at USD 69, 89 and 129 per user per year. The summary and watchOutFor record Basic 6.0.6, Visual 7.1.0 and Enhanced 7.0, the free and paid boundaries, the 14-day and 30-day trials and money-back terms, local processing and the absence of a hosted processing surface. The Basic platform list is now windows, mac and linux, the Visual platform list is windows, mac and linux, and Enhanced remains Windows-only; the editions record in-process processing with versions, licence models and trial terms. priceFromGbp stays 29 GBP per year as the Visual subscription and priceBasis stays year, and dataResidency stays unknown because the privacy policy permits storage in the US, Europe or elsewhere while desktop processing is local. The claims array, redaction, hasFreeTier and watermarkOnFree were checked and left unchanged.

  300. 7a053ac

    The PDFescape record, in the edit, protect, merge, compress, convert and sign jobs, the owner and licence fields, the summary, the watchOutFor and the two editions

    What it said
    The six jobs carried no scopes. The owner was the Avanquest brand rather than the Canadian contracting entity in the Terms of Use, and the licence field was empty. The summary and watchOutFor did not record the current Premium and Ultimate prices and billing basis, the current free and paid upload and history limits, the Desktop Windows requirements and Office dependency, or the absence of a published Desktop version number. The free online password route had no algorithm boundary and the Desktop-only merge, compression, conversion, certificate signing and redaction boundaries were not stated.
    What it says now
    The six jobs are scoped from the current first-party pages read on 2026-10-16. edit covers the free online annotation, text, shape, whiteout, image and basic form route and the paid Desktop editing of existing text and images, fonts, colours, highlighting, freehand, shapes, notes, bookmarks, links, stamps and watermarks, with no fidelity or layout guarantee and redaction kept separate. protect covers the free online password route without an algorithm claim and the Ultimate Desktop 256-bit encryption and permission route, with no reader-enforcement claim. merge covers the Desktop Premium/Ultimate merge, combine and extract operations with no preservation claim. compress covers the Desktop compression with no lossless guarantee. convert covers Desktop conversion to Word and other formats, creation from 300+ file types and print to PDF, with the Microsoft Office plug-in requirement and no fidelity guarantee. sign covers the Ultimate Desktop electronic and certificate signing and sealing, with no certificate-authority, trust-validation, timestamping or legal-validity guarantee. The owner is 7270356 Canada Inc. (Avanquest Software), the contracting entity under Quebec law, and the licence records the Avanquest Software brand, the EU controller Avanquest Software SAS, the proprietary freeware position, Free, Premium at USD 5.99 monthly or USD 2.99 per month billed yearly and Ultimate at USD 8.99 monthly or USD 5.99 per month billed yearly, the 15-day money-back guarantee, the 7-day Premium and Desktop trial, and the absence of an open-source licence. The summary and watchOutFor record the plans, prices, ten-megabyte and forty-megabyte limits, file-count and history limits, the Windows 7, 8 or 10 Desktop, the Office dependency, the Desktop-only advanced features, the no-watermark free service, and the absence of a published Desktop version number. The web edition records vendor-cloud processing with the US data-centre statement and plan-specific limits; the Windows edition records in-process Desktop processing with activation and Office requirements. priceFromGbp stays 4.44 per month as the current Premium monthly price at the recorded conversion, priceBasis stays month, dataResidency stays us and maxFileSizeMb stays 10 as the free online maximum. The claims array, redaction, hasFreeTier and watermarkOnFree were checked and left unchanged.

  301. 1130f1e

    The SumatraPDF record, in the view, convert, compress, merge, edit, protect and sign jobs, the owner and licence fields, the summary, the watchOutFor, the Windows edition and the redaction evidence

    What it said
    Six job tags had no scopes, and the merge scope was a one-line note rather than a documented scope. The owner field was empty although the site names Krzysztof Kowalczyk as the maker and the repository names the SumatraPDF Project Authors, and the licence recorded only GPLv3 without the BSD components, the (A)GPLv3 wording, the third-party notices or the absence of a corporate entity. The summary still said there was no redaction even though the 3.7 pre-release documents an applied redaction tool. The stable-versus-pre-release boundary was not applied across the conversion, compression, protection and signing tags, and the AI chat sidebar network exception was missing from the processing note.
    What it says now
    The seven jobs are scoped to the documented capabilities and version boundaries read on 2026-10-16. view covers the stable multi-format Windows reader for PDF, EPUB, MOBI, CBZ/CBR, DjVu, XPS, CHM and images with tabs, search, printing and read-aloud. edit covers the stable 3.3-plus annotation and markup editing of highlights, notes, stamps, images and ink saved back to the PDF, not body-text editing. convert covers the documented 3.7 pre-release tool converting PDF to images and grayscale, images, comics, EPUB and text to PDF and format selection by extension, with no fidelity guarantee. merge covers the 3.7 pre-release merge of pages from multiple files with no preservation claim. compress covers the 3.7 pre-release compression with no lossless or size guarantee. protect covers the 3.7 pre-release password encryption and decryption with no reader-enforcement or key-management guarantee beyond the password operation. sign covers the 3.7 pre-release certificate signing from the Windows certificate store or PFX/P12 with no certificate-authority, trust-validation, timestamping or legal-validity guarantee. The owner is Krzysztof Kowalczyk as the individual maintainer, and the licence records GPL-3.0 with BSD components, the (A)GPLv3/BSD README wording, the third-party notices and the absence of a corporate legal entity. The summary records the current stable 3.6.1, the 3.7 pre-release tool set, the 3.3-plus annotations, the absence of OCR and the AI chat sidebar network service, and no longer says there is no redaction. The watchOutFor records the Windows 7 to 11 32-bit, 64-bit and ARM64 builds, the pre-release tool boundaries, the unverified pre-release redaction, the certificate-signing boundary, the annotation boundary and the AI chat exception to local processing. The Windows edition records in-process processing with the version and network facts. The redaction evidence is re-read on 2026-10-16 and remains unverified and pre-release-only. priceFromGbp, priceBasis, pricingModel, hasFreeTier, platforms and dataResidency were checked and left unchanged, and the claims array was left unchanged.

  302. 8935866

    The PDFgear record, in the edit, convert, merge, compress, sign and ocr jobs, the owner and licence fields, the summary, the watchOutFor and the five editions

    What it said
    The six jobs carried no scopes. The owner field was empty although the privacy policy names PDF GEAR TECH PTE. LTD. in Singapore, and the licence recorded only proprietary without the free/future-paid position or the absence of a published price. The summary and watchOutFor treated the browser tools as uniformly cloud-based, although the merge and sign pages say they process locally in the browser, and they did not separate the visual electronic signature from the certificate-backed eSignGear product, record the OCR language figures or note the still-in-development searchable-PDF path. The browser edition processing was recorded as vendor_cloud without the per-tool split.
    What it says now
    The six jobs are scoped from the current first-party pages read on 2026-10-16. edit covers direct editing of existing text with font and size changes, existing text boxes and objects, images, text boxes, hyperlinks, shapes, highlights, stamps, watermarks, annotations, sticky notes, underlines, strike-through and forms, with no layout-fidelity guarantee. convert covers 60+ formats, Microsoft Office and image inputs, batch mode and page ranges, notes that some advanced conversions use cloud processing and makes no fidelity guarantee. merge covers joining PDFs, arranging files and pages with no watermark, records the local browser processing claim and treats without-losing-quality as marketing with no preservation claim. compress covers Low, Medium and Strong levels and the upload boundary, with no lossless or size guarantee. sign covers the visual electronic signature drawn, typed or uploaded and processed locally, and records that certificate-backed signing, audit logs and eIDAS/ESIGN workflows belong to the separate eSignGear product, with no certificate-authority or legal-validity guarantee for PDFgear. ocr covers editable Word, TXT and RTF recognition, 30+ languages for conversions and 10+ for area extraction, and the still-in-development searchable-PDF position, with no accuracy guarantee. The owner is PDF GEAR TECH PTE. LTD., 91 Bencoolen Street, #05-09 Sunshine Plaza, Singapore 189652, and the licence records the free proprietary position, the future-paid consideration and the absence of an open-source licence and published price. The summary and watchOutFor record the entity, the free status, the platform set, direct editing, 30+ OCR languages, the local and cloud split and the eSignGear separation. The browser edition processing is now unknown with the per-tool note that merge and sign claim local browser processing while compress and conversion upload and AI uses Azure OpenAI, and the installed editions remain in-process with the cloud conversion and AI exceptions. priceFromGbp stays null, priceBasis stays unknown, pricingModel stays free, hasFreeTier stays true, platforms stay windows, mac, ios, android and web, and dataResidency stays unknown because no hosting region is named. The claims array, redaction, watermarkOnFree and batchSupported were checked and left unchanged.

  303. f9bfcac

    The MuPDF.js record, in the edit, protect, merge, compress, view and extract jobs, the owner and licence fields, the summary, the watchOutFor and the two editions

    What it said
    Six jobs carried no scopes. The owner was the short brand rather than Artifex Software, Inc., and the licence did not record the AGPL-3.0-or-later package licence, the commercial per-copy/quarterly-minimum model or the npm package naming trap. The summary and watchOutFor did not record the current npm version 1.28.1 and its publication date, the absence of OCR and signing APIs, the browser and Node deployment split, the absence of a vendor-hosted endpoint or the unmeasured WebAssembly size. The editions had unknown processing states rather than browser in-process and self-hosted customer-server.
    What it says now
    The six jobs are scoped from the current docs and README read on 2026-10-16. view covers page.toPixmap rendering with matrix, colour space, transparency and annotation options and PNG/JPEG/PAM/canvas output, as rendering rather than an interactive viewer. edit covers creating and deleting annotations across the documented supported types including Redact, reading and writing form widget values, metadata and outlines, and page insert/delete/reorder/rotate/crop, with no general body-text rewriting claim and redaction left to its field. extract covers page.toStructuredText with plain and JSON output including blocks, lines, spans, positions and fonts, plus text search and links, with no OCR or image-extraction claim. merge covers PDFDocument graft/merge and page operations with no preservation claim. compress covers save-time garbage collection and compression with no lossless or size guarantee. protect covers opening password-protected PDFs, checking and setting permissions and encryption save options with no reader-enforcement or algorithm guarantee. The owner is Artifex Software, Inc., and the licence records AGPL-3.0-or-later or commercial, the Artifex dual-licensing page, the commercial per-copy and quarterly minimum model, the npm package naming and the absence of a published price. The summary records the npm package and version 1.28.1 and the absence of OCR and signing. The watchOutFor records the 1.28.1 release on 6 September 2026, the deprecated hyphenated package, browser and Node deployments, the absence of a vendor-hosted endpoint, the unmeasured WebAssembly size and the absence of OCR and signing. The web edition is in-process browser WebAssembly and the self-hosted edition is customer-server Node/Bun/Deno processing, both with sources. priceFromGbp stays null, priceBasis stays unknown, pricingModel stays open_source, hasFreeTier stays true, platforms stay web and self_hosted, interfaces stay library and browser, and dataResidency stays self_hosted. The redaction evidence remains the tested 1.28.1 result and the claims array was left unchanged.

  304. 4b99ff0

    The PDF-XChange Core API SDK record, in the edit, convert, merge, compress, sign and extract jobs, the owner and licence fields, the summary, the watchOutFor and the Windows edition

    What it said
    The six jobs carried no scopes. The owner was the Tracker Software Products brand rather than the parent PDF-XChange Co Ltd named on the site and licensing page, and the licence field did not record the England registration, the current version 11.1.0.0, the application-only sales notice, the evaluation watermark or the no-public-price position. The summary and watchOutFor did not distinguish the Core API column from the Editor SDK features, did not record the stale V8 label on the system-requirements page or the in-process DLL/COM deployment without a hosted endpoint, and the Windows edition had no processing note.
    What it says now
    The six jobs are scoped from the Core API column of the comparison chart and the SDK help manual read on 2026-10-16. edit covers creating and modifying PDFs and pages, placing objects, drawings, images and text, annotations, AcroForm creation and field data, bookmarks add/edit/reorder and crop/delete/reorder/add pages, with no existing-body-text rewrite claim. convert covers raster images plus EMF/WMF/AMF and .txt to PDF and export/render of pages and files to raster images, with no native Office or HTML conversion and no fidelity guarantee. merge covers merge, add, re-order, delete and crop page operations with no preservation claim. compress covers Zip, JPEG, JPEG2000, Run Length, LZW, JBIG2 and CCITT G3/4 options with no lossless or size guarantee. sign covers digital signatures, PFX/PFX2 and signature-field methods with no certificate-authority, trust, timestamp or legal-validity guarantee. extract covers text, image and AcroForm field-data extraction with no OCR claim. The owner is PDF-XChange Co Ltd, with Tracker Software Products recorded as a wholly owned subsidiary registered in England under number 11091579, and the licence records the distribution-volume tiers, the application-only sales notice, the evaluation watermark, the no-public-price position and the current version 11.1.0.0. The summary and watchOutFor record the current version, Windows 10 and later, the stale V8 label, the in-process DLL/COM deployment without a hosted endpoint, the Core API versus Editor SDK separation and the evaluation-watermark limitation. The Windows edition records in-process processing. priceFromGbp stays null, priceBasis stays unknown, pricingModel stays paid, hasFreeTier stays null, platforms stay windows, interfaces stay library, openSource stays false, selfHostable stays true and dataResidency stays unknown. The claims array was left unchanged.

  305. b9453f8

    The Scanner Pro record, in the ocr, view, merge, protect, convert and sign jobs, the licence field, the summary, the watchOutFor and the iOS edition

    What it said
    The six jobs carried no scopes. The licence recorded only proprietary without the Readdle Limited controller, the Readdle Technologies Limited App Store seller, the free-with-in-app-purchases model, the weekly/annual Plus subscription or the separate Fax Packs. The summary and watchOutFor did not record the current 9.0.2 version, the iPhone and iPad device family without a native Mac app, the OCR language bundles, the PDF-export-only password protection, the electronic rather than certificate-based signature, the PDF/JPEG/TXT export formats or the AI-provider retention boundary. The iOS edition had no processing note.
    What it says now
    The six jobs are scoped from the product page, App Store listing and support articles read on 2026-10-16. ocr covers the on-device OCR model that makes scans searchable, readable, selectable and copyable, with the Latin, Chinese+, Japanese+, Korean+, Cyrillic and Other language bundles, Latin preselected, and the Plus entitlement for new users, with no accuracy guarantee. view covers the interactive iPhone and iPad scanner and document viewer with search, sort, folders and full-text search after OCR, without a native Mac app claim. merge covers combining scans into one multi-page document and page reordering, without arbitrary external-PDF merging or preservation claims. protect covers the PDF export password that the recipient must enter to open the file and the app passcode, with no encryption-algorithm or reader-enforcement guarantee. convert covers the built-in converter and documented PDF, JPEG and TXT output, with no Office/Word conversion or fidelity claim. sign covers drawing a signature once and inserting it into scans and documents, with no certificate, validation, timestamp or CA claim. The owner stays Readdle Limited, company number 630281, Dublin 2, Ireland, as the controller named in the privacy notice, and the licence records the App Store seller Readdle Technologies Limited, the free download with in-app purchases, the 1-week and 1-year Scanner Pro Plus subscription and the separate Fax Packs, with no single price. The summary records version 9.0.2, the iPhone and iPad platform, the OCR bundles, PDF password protection, electronic signature and PDF/JPEG/TXT export. The watchOutFor records the no-single-price and subscription position, the free-tier limits, the Mac sync-not-native distinction, the OCR bundles and no accuracy figure, the PDF-export-only password protection, the electronic signature boundary, the no-file-storage position and the AI/fax network exceptions. The iOS edition records in-process scanning and OCR with the optional cloud sync, AI retention and fax boundaries. priceFromGbp stays null, priceBasis stays unknown, pricingModel stays freemium, hasFreeTier stays true, platforms stay ios, watermarkOnFree stays true, processesLocally stays true and dataResidency stays multi. The claims array was checked and left unchanged.

  306. 924b600

    The SwiftScan record, in the ocr, view, edit, merge, protect and sign jobs, the licence field, the summary, the watchOutFor and the iOS and Android editions

    What it said
    The six jobs carried no scopes. The licence recorded only proprietary without the Maple Media LLC US entity and address, the Maple Media Apps, LLC App Store seller, the October 2023 terms, the current VIP Monthly plans, the other Plus and Pro plans, the fax credits or the 7-day trial. The summary and watchOutFor did not record the current iOS 9.6.82 version, the absence of a native macOS or web app, the 60+ on-device OCR languages, the covers-content redaction limit, the Apple Intelligence requirement for the AI tools and their unconfirmed processing location, the PDF/app protection without encryption detail or the electronic-only signature. The iOS and Android editions had unknown processing states rather than in-process local scanning and OCR with the cloud, fax and AI exceptions.
    What it says now
    The six jobs are scoped from the product page, features page, store listings, privacy policy, terms and support articles read on 2026-10-16. ocr covers the on-device OCR with more than 60 languages, the Android automatic phone-language behaviour and the Plus/VIP entitlement, with no accuracy guarantee. view covers the iPhone, iPad and Android scanner and document viewer with search, folders and reading, without a native Mac or web app claim. edit covers Draw, Redaction hiding, Text addition, Notes, Signature, colour filters, rotation and page operations, without body-text rewriting. merge covers selecting multiple documents, reordering and saving the merged document, without preservation or arbitrary external-PDF claims. protect covers the app passcode/TouchID and the advertised PDF-scan protection, without encryption-algorithm or reader-enforcement claims. sign covers the electronic signature image, without certificate, validation, timestamp or CA claims. The owner stays Maple Media LLC, 9570 West Pico Blvd, Los Angeles, CA 90035 USA, and the licence records the App Store seller Maple Media Apps, LLC, the October 2023 terms, the free download with in-app purchases, SwiftScan VIP Monthly at GBP 5.99 in the UK and USD 5.99 in the US, the other Plus and Pro plans, the annual plans, one-off purchases, fax credits and the 7-day trial. The summary records version 9.6.82, the iOS and Android surfaces, 60+ on-device OCR languages, Plus/VIP, draw/text/notes/signature, merge, app/PDF lock, PDF/JPEG output, Apple Intelligence AI and the cloud/fax boundaries. The watchOutFor records that GBP 5.99 is the VIP Monthly plan rather than a general from-price, the other plans and free tier, the 7-day trial, OCR languages with no accuracy figure, the covers-content redaction limit, the Apple Intelligence requirement and unconfirmed AI processing location, fax and cloud as network paths, no native macOS or web app, password protection without encryption detail and the electronic-only signature. The iOS and Android editions record in-process scanning and OCR with optional cloud sync, fax and AI exceptions. priceFromGbp stays 5.99 with priceBasis month, platforms stay ios and android, dataResidency stays us, processesLocally stays true, hasFreeTier stays true and watermarkOnFree was checked and left unchanged. The claims array was checked and left unchanged.

  307. 82b1ecb

    The PDF Expert record, in the edit, sign, convert and compress jobs, the merge scope, the owner and licence fields, the price fields, the summary, the watchOutFor and the local and remote editions

    What it said
    The owner field was empty although the privacy notice and terms name Readdle Limited. The price recorded GBP 4.99 per week, which is not on the current pricing page and had been superseded by the Yearly GBP 72.99 plan and the Mac-only GBP 129.99 lifetime option. The summary and watchOutFor did not record the current iOS 7.26.2 and Mac 3.13.4 versions, the 7-day trial, the supported existing-body-text editing, the conversion formats and temporary three-hour server storage, the compression levels or the AI provider and retention details. The local and remote editions carried stale Windows, macOS, Linux and self-hosted platform lists that contradicted the Apple-only product, and the public prose saying there was no Windows, Android or web version created a false platforms-versus-prose candidate even though the platform list was correct.
    What it says now
    The owner is Readdle Limited, company number 630281, Glandore Business Centres, 26-27 Fitzwilliam Place, Fitzwilliam Hall, Dublin 2, D02T292, Ireland, as named in the privacy notice and terms, and the licence records Readdle Limited as controller, Readdle Technologies Limited as the App Store and Mac App Store seller, the Yearly GBP 72.99 plan and the Mac-only Lifetime GBP 129.99 option, the 30-day money-back guarantee, the 7-day trial and the absence of an open-source licence. The price is now 72.99 with basis year, from the current GBP pricing page, and the separate USD store prices are recorded as separate without conversion. The summary records the Apple-only product, the current versions 7.26.2 and 3.13.4, the editing, conversion, compression, signing, OCR, AI and trial facts, and the watchOutFor records the Yearly and Lifetime plans, the limited free tier, body-text editing, conversion formats and temporary three-hour server storage, the Lossless High and Medium compression levels with no percentage guarantee, the electronic-only signature, the AI providers and 30-day retention, and the Apple-only distribution. The local and remote editions now list mac and ios, with installed editing in-process and conversion, AI and cloud sync in the vendor-cloud edition, preserving the Mixed state. The public prose no longer uses Windows, Android or web wording and the merge scope no longer mentions those platforms. The edit, sign, convert and compress jobs are scoped from the current official pages, the merge and OCR scopes remain, and the platforms-versus-prose candidate is cleared. platforms stay mac and ios, pricingModel stays freemium, hasFreeTier stays true, dataResidency stays multi, processesLocally stays true, and the claims array was checked and left unchanged.

  308. cd9279c

    The Nitro PDF Pro record, in the edit, sign, convert, ocr and protect jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the local and remote editions

    What it said
    The owner field was empty although the privacy policy and footer name Nitro Software, Inc. The price recorded GBP 138 per year, which is not on the current rendered pricing page; that page now publishes USD 15 per user/month billed annually at USD 180.00 and Classic at USD 270.00 per licence billed once for a three-year Windows term, with no GBP figure. The summary and watchOutFor did not record the current macOS 26.4 and iOS 26.2 versions, the separate Nitro Sign/Smart Redact/Automate/AI cloud services, the certificate-based digital-signature boundary, the OCR language and quality settings or the document encryption levels. The local and remote editions carried stale Windows, macOS, Linux and self-hosted platform lists that contradicted the Windows/macOS/iOS product.
    What it says now
    The owner is Nitro Software, Inc., 447 Sutter St, STE 405 #1015, San Francisco, CA 94108, USA, and the licence records the current Standard, Plus and Classic plans, the USD 180.00 annual Standard price, the USD 270.00 three-year Classic price, the 14-day trial and the absence of an open-source licence. The stale GBP 138 per year price is removed, priceFromGbp is null and priceBasis is unknown because the current page publishes USD and no GBP figure, and no currency conversion is performed. The summary records the macOS 26.4 and iOS 26.2 versions, the Windows/macOS/iOS surfaces, the local editing/OCR/export path and the separate Nitro Sign/Smart Redact/Automate/AI cloud services. The watchOutFor records the current USD plans and trial, the no-GBP position, the digital ID/certificate signature boundary with no CA, trust or timestamp guarantee, the separate Nitro Sign service, the OCR language and quality settings, the 40-bit RC-4/128-bit AES/256-bit AES levels and the no reader-enforcement guarantee, and the absence of a current first-party Apryse ownership statement. The local and remote editions now list Windows, macOS and iOS, with installed editing, OCR and export in-process and Nitro Sign/Smart Redact/Automate/AI in the vendor-cloud edition, preserving the Mixed state. The five jobs are scoped from the current user guides and product pages. platforms stay windows, mac and ios, pricingModel stays paid, hasFreeTier stays false, dataResidency stays unknown and processesLocally stays true. The claims array was checked and left unchanged.

  309. aaffaa1

    The Nitro PDF Pro licence, summary and watchOutFor seller/billing provenance

    What it said
    The licence recorded Nitro Software, Inc. as the seller and biller of the subscription and Classic licences without distinguishing the direct-checkout seller. The current Nitro Ecommerce Terms name Nitro Software Ecomm Limited, First Floor, 80 Harcourt Street, Dublin, Dublin 2, D02F449, Ireland, as the legal entity selling and billing for licenses through checkout.gonitro.com, acting as an Authorized Reseller to Nitro Software Inc, and say Nitro (not Nitro Ecomm) provides and operates the services. The App Store listings separately name Nitro Software, Inc. as seller, so the direct-checkout and App Store seller surfaces needed to be recorded separately. The owner/controller, the USD prices, the five job scopes and the counters were already correct and were not changed.
    What it says now
    The owner remains Nitro Software, Inc., 447 Sutter St, STE 405 #1015, San Francisco, CA 94108, USA, as the controller named in the privacy policy and site footer. The licence records Nitro Software Ecomm Limited, First Floor, 80 Harcourt Street, Dublin, Dublin 2, D02F449, Ireland, as the seller and biller for licences purchased through checkout.gonitro.com and as an Authorized Reseller to Nitro Software Inc.; it records that Nitro, not Nitro Ecomm, provides and operates the services; and it records the App Store listings separately as naming Nitro Software, Inc. as seller. The summary and watchOutFor state which seller covers the direct-checkout and App Store surfaces, the USD 180 annual Standard and USD 270 Classic prices remain attributed to the direct-checkout surface, priceFromGbp stays null and priceBasis stays unknown with no currency conversion, and all five job scopes and editions are unchanged.

  310. 3ba27f5

    The Spire.PDF record, in the edit, convert, protect, sign and extract jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the four editions

    What it said
    The five jobs carried no scopes and the owner field was empty because no legal entity had been captured. The price recorded GBP 747.57 as a one-off, but that was a synthetic conversion of a USD 999 vendor price and not a vendor-published GBP figure. The summary and watchOutFor did not record the current package version 12.10.0, the target frameworks, the Word converter documentation reporting an incomplete paragraphs-and-tables limitation, the absence of OCR, the separate Chengdu signing identity named in the package certificate, or the blocked current pricing page. The four editions had unknown processing states rather than in-process local library execution.
    What it says now
    The five jobs are scoped from the current NuGet package 12.10.0 XML API documentation. edit covers PdfTextReplacer and text widgets, annotations, forms, pages, bookmarks, attachments and overlays, with no layout-fidelity guarantee. convert covers the documented PdfToWordConverter, PdfToDocConverter, PdfToImageConverter, PdfToSvgConverter, PdfToMarkdownConverter, PdfStandardsConverter, OfdConverter, PdfToLinearizedPdfConverter, PdfGrayConverter and PdfCompressor APIs, with the Word converter documentation reporting an incomplete paragraphs-and-tables limitation, no Excel or PowerPoint converter claim and no fidelity guarantee. protect covers PdfSecurity/PdfPasswordSecurityPolicy, 40-bit, 128-bit and 256-bit keys and PdfPermissionsFlags, with no reader-enforcement guarantee. sign covers PdfSignature and PdfCertificate from X509 and PFX, with no CA, trust-validation or timestamp guarantee. extract covers PdfTextExtractor, PdfImageHelper, PdfTableExtractor, forms and attachments, with no OCR or accuracy claim. The owner is e-iceblue Corporation as the EULA licensor; the licence records the New York law EULA, the evaluation, developer, developer OEM, site enterprise and site OEM grants, the OEM restriction for public-facing, Docker and cloud deployments, and the separate Chengdu E-iceblue Technology Co., Ltd signing identity named in the package certificate, with the relationship and street address unverified because the vendor site was blocked. priceFromGbp is null and priceBasis is unknown because GBP 747.57 was a synthetic conversion of USD 999 and the current vendor pricing page was Cloudflare-blocked. The summary records Spire.PDF 12.10.0, the net40/net48/net6/net9/net10 assemblies, the in-process Windows/Linux/macOS library model, the documented edit/convert/sign/protect/extract APIs and the separate Java edition not audited. The watchOutFor records the price conversion removal and blocked page, the free-edition limits as previously documented, the incomplete Word conversion, the absence of OCR, the certificate and encryption boundaries, the signing-identity conflict and the no-hosted-endpoint position. All four editions are in_process local execution with no vendor endpoint. platforms stay windows, linux, mac and self_hosted, interfaces stay library, pricingModel stays paid, hasFreeTier stays true, openSource stays false, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  311. ea16300

    The No Fuss PDF record, in the edit, sign, merge, convert and extract jobs, the licence field, the summary, the watchOutFor and the web edition

    What it said
    The five jobs carried no scopes and the licence field was blank even though the current site publishes no licence or terms page, which is itself a finding rather than a missing value. The summary said the tools extract pages but did not distinguish page extraction from text, image, table or form extraction, and the watchOutFor did not record the absent licence and terms, the absent registered address, company number, jurisdiction, version and file limits, or the page-extraction meaning of the extract job. The web edition had no processing note for the browser-local PDF.js and pdf-lib path or the same-origin telemetry.
    What it says now
    The five jobs are scoped from the current tool pages read on 2026-10-16. edit covers the Fill PDF visible text, date, check, cross and signature overlays on pages including scanned and flattened forms, with no interactive-field creation, no original-text change and no body-text rewrite claim. sign covers the visible drawn, typed or uploaded PNG or JPEG signature, with an explicit statement that it is not a certificate-backed digital signature. merge covers combining two or more PDFs in the chosen order in the browser. convert covers JPG, JPEG and PNG images to one PDF and PDF pages to JPG images, with a single JPG or ZIP output and a quality choice for PDF to JPG. extract covers Split PDF page extraction and split-every-page into a ZIP, and explicitly excludes text, image, table and form-data extraction. The licence field records that no licence or terms page is published, that the sitemap lists only the tools, Privacy, About, Compare and Research, and that no open-source licence or source distribution is offered. The summary records the six tools, the browser-local PDF.js and pdf-lib processing and the free no-account no-upload no-watermark position. The watchOutFor records the visual-signature boundary, the absence of body-text editing, interactive-form creation, OCR and redaction, the page-extraction meaning of extract, the JPG and PNG conversion limits, the absent licence, terms, address, company number, jurisdiction, version and file limits, and the same-origin analytics exception. The web edition records in-process browser processing with the telemetry note. platforms stay web, pricingModel stays free, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, dataResidency stays unknown and processesLocally stays true. The claims array was checked and left unchanged.

  312. a41a7bc

    The PDF-XChange Editor record, in the edit, convert, ocr, sign and compress jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the local and remote editions

    What it said
    The owner was the Tracker Software Products brand rather than PDF-XChange Co Ltd, the parent named on the product and licensing pages. The price recorded GBP 47.00 and the version 11.0.1.0, while the current pages give GBP 48.00 and version 11.1.0.0; the licence field did not record the per-user perpetual model, the included first year of maintenance or the no-refund position. The summary and watchOutFor did not record the 402 feature count with 257 free and 145 licensed, the Editor Plus enhanced OCR and form features, the PAdES signing and long-term validation capabilities, or the AI and cloud-storage exceptions. The local and remote editions carried stale Windows, macOS, Linux and self-hosted platform lists that contradicted the Windows-only product.
    What it says now
    The owner is PDF-XChange Co Ltd, with Tracker Software Products recorded as a wholly owned subsidiary registered in England under number 11091579. The licence records the per-user perpetual model, the Single User Editor at GBP 48.00 plus VAT and Editor Plus at GBP 61.00 plus VAT, the included first year of maintenance, the no-refund position after licence keys are issued, the conditional evaluation watermark and the Editor Plus enhanced OCR, fillable forms, dynamic stamps, comparison, Identify Forms and Add Placeholder features. The price is 48 with basis one_off, and the summary and watchOutFor record version 11.1.0.0, Windows 10 and later, the 402/257/145 feature counts, the approximate 70 per cent free claim, the PAdES digital signature, certification, timestamp and long-term-validation path, the separate visual signature and FreeSign path, the default OCR in Editor versus Enhanced OCR in Editor Plus, the documented conversion formats, the Save as Optimized compression and the optional AI and cloud-storage network paths. The local edition now lists Windows only with in-process processing, and the remote edition lists Windows only with vendor-cloud processing for the optional AI and cloud-storage integrations. The five jobs are scoped from the current feature list. platforms stay windows, pricingModel stays freemium, hasFreeTier stays true, dataResidency stays unknown and processesLocally stays true with the AI/cloud exceptions. The claims array was checked and left unchanged.

  313. 986553c

    The PDF Studio record, in the edit, sign, view, convert and ocr jobs, the owner and licence fields, the website and price fields, the summary, the watchOutFor and the three editions

    What it said
    The owner was the vague brand Apryse rather than the current product EULA licensor Apryse Software Corp. The price recorded GBP 14.83 per month as a synthetic conversion of USD 19.99, and the website still pointed at the legacy Qoppa product page. The summary and watchOutFor did not record the current Xodo PDF Studio 2026.0.0 version, the USD-only monthly, yearly and perpetual options, the watermarked demo and credit-card trial, the rebrand from Qoppa, the separate PDF Studio Viewer and Xodo Sign add-on, the OCR language documents or the current signing, viewing and conversion boundaries. The three editions had unknown processing states rather than local desktop execution.
    What it says now
    The owner is Apryse Software Corp., 530 Harrison Ave, Second Floor, Boston, MA 02118 USA, as the licensor named in the current Xodo PDF Studio EULA last modified 21 November 2025 and governed by Delaware law. The licence records the subscription and perpetual grants, the USD 19.99 monthly, USD 9.99 per month yearly and USD 240 per-user perpetual prices, the volume discounts, the watermarked demo and credit-card trial, and the unresolved first-party naming conflict between Apryse Software Corp. in the product EULA, Apryse Software Inc. in the Xodo terms and privacy notice, and Qoppa Software in the legacy EULA. The website is now https://xodo.com/pdf-studio. priceFromGbp is null and priceBasis is unknown because the vendor publishes USD only and the old GBP 14.83 was a conversion; no conversion is used. The summary records Xodo PDF Studio 2026.0.0 released 31 March 2026, the Windows, macOS and Linux platforms, the local offline processing, the editing, signing, viewing, conversion, OCR, redaction, sanitisation, optimisation and accessibility features, and the separate PDF Studio Viewer and Xodo Sign add-on. The watchOutFor records the entity conflict, the USD-only pricing, the trial and demo conditions, the rebrand, the OCR common languages and the absence of a fixed total or accuracy figure, the digital signature and certification boundary, and the local desktop processing with the separate add-on service. The edit, sign, view, convert and OCR jobs are scoped from the current Xodo feature comparison and the PDF Studio knowledge base, and the previously verified merge, protect, compress and accessibility scopes were not reopened. All three editions are in_process local desktop execution. platforms stay windows, mac and linux, pricingModel stays paid, hasFreeTier stays false, dataResidency stays unknown and processesLocally stays true. The claims array was checked and left unchanged.

  314. e238cfc

    The MuPDF mutool record, in the convert, merge, sign, edit and protect jobs, the owner and licence fields, the summary, the watchOutFor and the five editions

    What it said
    The owner was the short name Artifex Software rather than Artifex Software, Inc., and the licence field did not record the AGPL-3.0-or-later or commercial position with the Artifex address and the absence of a published price. The summary and watchOutFor did not record the current stable 1.28.5 tag or the 1.29.0 development header, the exact mutool command set, the convert output-option families, the certificate-based sign workflow, the clean encryption and permission controls, the absence of a dedicated redaction subcommand or the draw OCR options. The five editions had unknown processing states rather than local command-line and library execution.
    What it says now
    The owner is Artifex Software, Inc., 39 Mesa Street, Suite 108A, San Francisco, CA 94129, USA, and the licence records AGPL-3.0-or-later or commercial from Artifex, with no published commercial price and a commercial route through Artifex sales. The summary records the current stable 1.28.5 release tag dated 25 September 2026 and the 1.29.0 development header in master as an unreleased version, the mutool command set, the convert output families, merge, clean, create, pages, draw, extract, sign and run, and the local CLI and library model. The watchOutFor records the toolkit-not-editor position, the absence of a dedicated redaction subcommand, the draw OCR options without an OCR job or accuracy guarantee, the certificate-based sign and verify and clear workflow without CA or trust or timestamp guarantees, the rc4-40, rc4-128, aes-128 and aes-256 encryption with owner and user passwords and permissions without reader-enforcement guarantees, and the absence of any hosted or cloud service. The convert, merge, sign, edit and protect jobs are scoped from the man page, and the previously verified compress scope was not reopened. All five editions are in_process local execution, with api and self_hosted recorded as library or local-build surfaces. platforms stay windows, mac, linux, api and self_hosted, interfaces stay library and cli, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  315. 5a9ea9a

    The pikepdf record, in the protect, merge and edit jobs, the owner and licence fields, the summary, the watchOutFor, the platform prose and the four editions

    What it said
    The owner field was empty although PyPI and the documentation name James R. Barlow as author and copyright holder. The licence field recorded only MPL-2.0 and did not distinguish pikepdf from the separate qpdf Apache-2.0 project with its own copyright holders and bundled third-party notices. The summary and watchOutFor did not record version 10.17.0, the Python 3.11 or later requirement, the exact wheel platforms and the macOS Apple Silicon-only limit, the local in-process library model, or the qpdf permission-enforcement caveat. The public prose said web delivery and then browser surface, which created a false platforms-versus-prose candidate although the platform list is the in-process library platforms. The four editions had unknown processing states.
    What it says now
    The owner is James R. Barlow, the PyPI author and 2026 documentation copyright holder. The licence records MPL-2.0 for pikepdf, Apache-2.0 separately for qpdf with the 2005-2021 Jay Berkenbilt and 2022-2026 Jay Berkenbilt and Manfred Holger copyright, and the bundled third-party notices, with no corporate ownership relationship asserted. The summary records pikepdf 10.17.0, Python 3.11 or later, the documented wheel table for Python 3.11 through 3.14 on Windows x64 and ARM64, Linux x86_64 and aarch64 on manylinux and musllinux, and macOS 15+ on Apple Silicon, with Intel macOS source-build only. The watchOutFor records the library-only surface, the no-rendering/no-text-extraction/no-OCR/no-conversion limits, the redaction limit, the encryption and permission API and the qpdf caveat that restrictions are enforced only by the opening software and can be ignored, the separate qpdf licence and copyright, the absence of a public commercial price or support price, and the local in-process model with no mobile or hosted surface. The protect, merge and edit jobs are scoped from the current documentation, and the previously verified compress scope was not reopened. All four editions are in_process local execution. platforms stay windows, mac, linux and self_hosted, interfaces stay library, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier is true for the MPL library, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  316. 602a8d0

    The SimplePDF record, in the edit, merge, sign and convert jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the web, hosted and api editions

    What it said
    The owner field was empty and the licence field was empty. The price recorded GBP 73.43 per month as a synthetic conversion of USD 99. The summary and watchOutFor did not record the live operator change from Tree Social O� in Estonia to SimplePDF B.V. in the Netherlands in October 2026, the Netherlands governing law, the current USD plan set, the free-editor versus paid server-side processing and storage split, the Premium API access, the separate MIT embed SDK or the PDF-to-web-form conversion scope. The editions had unknown processing states and the API edition was marked as not requiring an account.
    What it says now
    The owner is SimplePDF B.V., Pieter Baststraat 24H, 1071 TX Amsterdam, Netherlands, the operator named in the live privacy policy and terms last updated 15 October 2026, whose version history says it replaced Tree Social O� in October 2026 and that Netherlands law replaced Estonian law. The licence records that no software licence is published for the hosted service, that the service is governed by its Terms and Conditions, and that the separate SimplePDF/simplepdf-embed SDK is MIT-licensed without applying that licence to the hosted service. priceFromGbp is null and priceBasis is unknown because the vendor publishes USD only and the old GBP 73.43 was a conversion; the current Free, Basic USD 99, Pro USD 349 and Premium USD 899 monthly plans, the Premium API access and the unavailable annual amounts are recorded without conversion. The summary and watchOutFor distinguish the client-side free editor from the server-side paid dashboard, submissions, notifications, storage without BYOS and API paths, record the BYOS path, the visual signature boundary, the PDF-to-web-form conversion scope, the absence of OCR, compression, password protection and redaction, the MIT SDK separation and the unverified self-serve API status. The four jobs are scoped from the current feature, signature and PDF-to-web-form pages. The editions are the free browser editor in_process without an account, the paid workflows vendor_cloud with an account, and the Premium API vendor_cloud with an account, producing the Mixed state. platforms stay web and api, interfaces stay http and browser, pricingModel stays freemium, hasFreeTier stays true, dataResidency stays eu, processesLocally stays true for the free editor with the paid server exceptions recorded, and redaction stays none. The claims array was checked and left unchanged.

  317. 353b6a6

    The PDFsharp record, in the edit, protect, merge and convert jobs, the owner and licence fields, the summary, the watchOutFor, the version and target notes and the four editions

    What it said
    The owner did not include the Troisdorf location, the licence was recorded as MIT, and the tagline and bestFor described a permissively licensed library. The summary and watchOutFor referred to the older 6.2 line and did not record the current 6.5.0 stable and 7.0.0 Preview 2 releases, the AGPL-3.0-only or commercial licensing, the Core net8/net9/net10/netstandard2.0 targets or the Windows-only GDI/WPF targets. The internal notes still claimed MIT as verified and left hasFreeTier null. The four editions had unknown processing states.
    What it says now
    The owner is empira Software GmbH, Troisdorf (Cologne Area), Germany. The licence is AGPL-3.0-only or commercial, with the NuGet packages for PDFsharp 6.5.0 and 7.0.0-preview-2 and PDFsharp-MigraDoc 6.5.0 declaring AGPL-3.0-only and the docs, README and support page describing the AGPL and commercial options. The earlier MIT value is recorded as historical and superseded. The summary and watchOutFor record PDFsharp 6.5.0 stable and 7.0.0 Preview 2, both dated 2026-10-13, the Core targets net8.0, net9.0, net10.0 and netstandard2.0, the GDI and WPF targets net8.0-windows7.0, net9.0-windows7.0, net10.0-windows7.0 and net462, MigraDoc PDF/RTF generation, the absence of a renderer, text extractor and existing-text editor, the unverified general forms and metadata position, the Core FontResolver and image limits, the encryption methods and permissions, and the certificate-backed signatures that are documented but not added as a sign job. The edit, protect, merge and convert jobs are scoped from the current feature, encryption and MigraDoc pages; convert covers PDF creation and MigraDoc PDF/RTF output only. All four editions are in_process local library execution, with Windows covering Core/GDI/WPF and macOS and Linux using Core only. platforms stay windows, mac, linux and self_hosted, interfaces stay library, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier is true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  318. e04cd3e

    The Pdftools SDK record, in the convert, compress and sign jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the four editions

    What it said
    The owner was Apryse rather than the legal provider PDF Tools AG, and the licence wording described a usage-based yearly licence rather than the current per-page, quote-based model with watermarked unlicensed and trial output and page credits. The version notes did not record the live 1.21.0 release, the summary and watchOutFor did not distinguish the separate OCR Service or the licensing service and gateway from document processing, and the platform notes did not record the current language, operating-system and ARM64 details. The four editions had unknown processing states.
    What it says now
    The owner is PDF Tools AG, Steinstrasse 21, 8003 Zurich, Switzerland, the legal provider named in the imprint and licence agreement. The site banner saying Pdftools is joining Apryse is recorded as a prospective corporate transition, not used to replace the provider. The licence records the per-page, quote-based commercial model, the absence of a public numeric rate, the watermarked unlicensed and trial output, the page-credit counts, and the distinction between the licensing service or gateway and document processing. priceFromGbp is null and priceBasis is unknown, and hasFreeTier is false because unlicensed and trial output is watermarked. The summary and watchOutFor record the live 1.21.0 release dated 13 October 2026, the earlier 1.20.1 and 1.20.0 releases, the C/C#/Java/Python and Go bindings, the Windows x86/x64, macOS x64/ARM64 and Linux x64 glibc platforms plus the ARM64 release-note additions, the separate OCR Service, the image-to-PDF and PDF/A and PDF-to-image conversion profiles, the Optimize image recompression and MRC profiles, the PFX/P12 certificate-backed signing with certification, timestamp, LTV and validation, and the absence of a hosted conversion API or CA/trust guarantee. The convert, compress and sign jobs are scoped from the current guides. All four editions are in_process local/native execution, with self_hosted describing customer-operated deployment. platforms stay windows, linux, mac and self_hosted, interfaces stay library, pricingModel stays paid, openSource stays false, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true with the licensing and OCR-service exceptions recorded. The claims array was checked and left unchanged.

  319. 14b104a

    The SignNow record, in the sign, edit and convert jobs, the owner and licence fields, the price fields, the summary, the watchOutFor and the four editions

    What it said
    The owner was the unqualified airSlate brand rather than the contracting-entity rule and provider list. The licence field was empty. The price recorded GBP 14.83 per month as a synthetic conversion of USD 8, and the pricing model was described as invite-based without the current user-plan and Site License structure. The summary and watchOutFor did not record the hosted server-side processing and US transfers, the 13 October 2026 privacy date, the 2 March 2023 terms date, the current USD plans, the Site License API entitlement, the absence of a permanent free tier, the retention position, the SES/AES/QES plan levels or the document-generation meaning of the convert job. The four editions had unknown processing states.
    What it says now
    The owner is airSlate, Inc., with the Contracting Entity Table rule recorded in the licence and watchOutFor: the applicable provider is determined by services, location and payment method and is listed in the Subscription Terms. The provider list includes airSlate, Inc. (Delaware, notice address 17 Station Street, Ste. 203, Brookline, MA), airSlate Legal Forms, Instapage, DocHub, airSlate Ireland Limited, airSlate Poland and airSlate Australia, each with its own law and venue. The licence records the Terms of Service last updated 2 March 2023 and the Subscription Terms. The privacy notice last updated 13 October 2026 lists uploaded Documents and Electronic Signatures, billing and usage information, says services will not function without transfers to the United States, and gives retention purposes rather than a fixed duration. priceFromGbp is null and priceBasis is unknown because the vendor publishes USD and the payment page rendered only a JavaScript shell; the old GBP 14.83 conversion is removed and the current Business, Business Premium, Enterprise, Site License and Business Cloud USD plans and 7-day trial are recorded without conversion. hasFreeTier is false because only a 7-day trial was found. The summary and watchOutFor record the hosted server-side service, US data residency and transfers, the web, iOS, Android and API surfaces, the API Site License entitlement, the electronic-signature workflow and SES/AES/QES plan claims without a certificate/CA/timestamp or universal legal guarantee, and the document-generation meaning of convert. The edit, sign and convert jobs are scoped from the developer documentation and airSlate information page. All four editions are vendor_cloud hosted with accounts, platforms stay web, ios, android and api, interfaces are http and browser, pricingModel stays paid, dataResidency stays us, processesLocally is false and openSource stays false. The claims array was checked and left unchanged.

  320. 69c7040

    The PandaDoc record, in the sign, edit and convert jobs, the owner and licence fields, the price fields, the free-plan claim, the summary, the watchOutFor and the web, ios, android and api editions

    What it said
    The owner field was empty although the legal pages name PandaDoc, Inc. The licence field recorded proprietary without the hosted MSA/DPA/order-form model. The price recorded GBP 25.96 per month as a synthetic conversion of USD 35, and the summary and watchOutFor described the free tier as a flat 60 documents per year, which is now only the older annual Free account allowance; new Free accounts send 5 documents per month. The record did not record the hosted server-side processing, the US/EU data-center choice, the 31 July 2026 privacy date, the Free API eligibility after 5 June 2026, the production API entitlement, the sandbox limits, the mobile apps or the document-generation meaning of convert. The four editions had unknown processing states.
    What it says now
    The owner is PandaDoc, Inc., 548 Market St PMB 185308, San Francisco, CA, 94104-5401, US, and the licence records the hosted Master Services Agreement, Data Processing Agreement and order-form model rather than a software licence. The summary and watchOutFor record the hosted SaaS processing, the US/EU data-center choice with the EU instance in Frankfurt, the privacy notice updated 31 July 2026, and the current plan set: Free eSign USD 0, Starter USD 19 annual (USD 35 monthly), Business USD 49 annual (USD 65 monthly) and Enterprise by quote. The Free plan is recorded as 5 documents per month for new accounts and 60 per year for older annual accounts, with 2 recipients, 5 templates, unlimited storage/users/fields; API access is limited to Free accounts created after 5 June 2026 and does not cover OAuth app creation or webhooks, while production API access requires a plan with API entitlement. The price is null with priceBasis unknown because the vendor publishes USD and no GBP figure, and the old GBP 25.96 conversion is removed. The first claim is refreshed to the current 5-per-month and older-annual 60-per-year wording with a new evidence artefact and hash. The edit job covers templates, PDFs, DOCX, fields/placeholders, content/quote changes, roles, routing, collaboration, version history and embedded editing; sign covers typed/drawn signatures, recipient verification, signing order, audit and signed-PDF retrieval with ESIGN/UETA/eIDAS statements and QES as a paid add-on, without certificate/CA/timestamp or universal legal guarantee; convert is document generation from templates, PDFs, DOCX and CRM/data into signature-ready documents, not generic conversion. Platforms now list web, ios, android and api; interfaces are http and browser; all four editions are hosted vendor_cloud with accounts; pricingModel stays freemium, hasFreeTier is true, dataResidency stays multi, processesLocally is false and openSource stays false.

  321. a797eb7

    The qpdf record, in the merge, compress and protect jobs, the owner and licence fields, the summary, the watchOutFor and the three editions

    What it said
    The owner field was empty although the README identifies Jay Berkenbilt and Manfred Holger as the copyright holders. The summary and watchOutFor did not record the current 12.4.2 release, the exact binary matrix, the C++20 build requirement, the C/C++ APIs and QPDFJob, the JSON structural interface or the encryption restrictions. The public summary used the phrase web viewing, which created a false platforms-versus-prose candidate although the platform list is a local CLI/library platform list. The three editions had unknown processing states.
    What it says now
    The owner is Jay Berkenbilt and Manfred Holger, with the README copyright of 2005-2021 Jay Berkenbilt and 2022-2026 Jay Berkenbilt and Manfred Holger. The licence is Apache-2.0 and the current stable release is qpdf 12.4.2, published 27 September 2026. The summary and watchOutFor record the official macOS Intel/Apple Silicon, Linux x86_64/aarch64 and Windows 32/64/ARM64 binaries plus source and documentation, the C++20 build requirement as a build rather than runtime claim, the C/C++ APIs and QPDFJob, and the JSON structural inspection/update path. The merge job covers --pages, --collate and --split-pages page selection/merge/split/collate; the compress job covers --compress-streams, --decode-level, --recompress-flate, --object-streams, --optimize-images, --jpeg-quality, --linearize and JSON structural operations, without a lossless or size guarantee; the protect job covers user and owner passwords, RC4/AES options, permissions, decryption, copy-encryption and remove-restrictions, with the qpdf caveat that readers may ignore restrictions and that 40-bit/default RC4 are insecure. No content editing, rendering, text extraction, OCR, redaction, signing or format conversion is claimed, and pikepdf is kept separate. The public prose no longer contains web or browser wording. All three editions are in_process local CLI/library execution; platforms stay windows, mac and linux, interfaces stay library and cli, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  322. 5921105

    The Poppler record, in the extract, view and merge jobs, the owner and licence fields, the summary, the watchOutFor and the four editions

    What it said
    The owner field was empty although the source headers carry the Poppler Developers copyright. The summary and watchOutFor did not record the current 26.10.0 release, the GPLv2/GPLv3 licence split, the official CI platform split, the exact utility surface or the pdfsig verification and 26.10 CAdES-B signing capability. The extract, view and merge jobs had no scopes. The four editions had unknown processing states.
    What it says now
    The owner is The Poppler Developers. The licence records GPL-2.0-or-later for Poppler project changes over the xpdf-3.0 GPLv2 base, with COPYING GPLv2 and COPYING3 GPLv3 for bundled components and no permissive option. The current stable release is 26.10.0, released 4 October 2026. The summary and watchOutFor record the C++/GLib/Qt5/Qt6 rendering frontends, the pdftotext/pdfimages/pdfinfo/pdffonts/pdfdetach extraction utilities, the pdftoppm/pdftocairo rendering utilities, pdfunite/pdfseparate merge and split, and pdfsig signature verification and signing. The official CI covers Debian unstable, Ubuntu 24.04, Fedora clang and mingw64 Windows cross-build and Qt 6 Android; native macOS and native Windows are community-supported. The extract job covers text, image, metadata, font and attachment extraction with no OCR or arbitrary text editing; the view job covers library/frontend rendering and raster/vector utilities with no finished GUI viewer or exact-rendering guarantee; the merge job covers pdfunite concatenation and pdfseparate splitting, with the pdfunite encrypted-source limitation. Signature verification and signing are recorded separately, including 26.10 CAdES-B/eIDAS support, with no CA, trusted timestamp, universal trust or legal-validity guarantee and no sign job added. The convert job was already verified and was not reopened. All four editions are in_process local library/CLI execution; platforms stay linux, windows, mac and android, interfaces stay library and cli, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  323. c4b7664

    The pypdf record, in the extract, merge, edit and protect jobs, the owner and licence fields, the summary, the watchOutFor and the four editions

    What it said
    The owner field was empty although the project is collaborative under py-pdf and the pyproject names the author and maintainers. The summary and watchOutFor did not record the current 6.21.1 release, the PyPI 6.21.1 package, the Python >=3.9 and 3.9-3.15 support, the py3-none-any wheel, the BSD-3-Clause author/maintainer provenance or the exact extraction, merging, form/annotation and encryption boundaries. The extract, merge and protect jobs had no scopes, and the record did not state that there is no convert job or generic conversion. The four editions had unknown processing states.
    What it says now
    The owner is pypdf contributors and the licence records BSD-3-Clause with copyright 2006-2008 Mathieu Fenniak and other contributors, the pyproject author Mathieu Fenniak and maintainers stefan6419846 and Martin Thoma, and the collaborative py-pdf structure. The current release is 6.21.1, published 17 October 2026 after 6.21.0 on 16 October 2026; PyPI already serves 6.21.1 and is not lagging. The package requires Python >=3.9, declares support for 3.9 through 3.15 and ships a py3-none-any wheel, so no OS-specific binary or hosted service is claimed. The extract job covers PdfReader/page.extract_text text-layer, layout and visitor extraction, metadata, images and attachments without OCR; merge covers PdfWriter.append/merge, add_page/insert_page, page reordering/selection and form grouping; edit covers page rotation/cropping/transformation, form filling/flattening, annotations, stamps, metadata, outlines and PDF serialisation; protect covers PdfWriter.encrypt/PdfReader.decrypt with RC4-40, RC4-128, AES-128, AES-256-R5 and AES-256, the RC4 default warning and the AES extra dependency. The catalog has no convert job and no generic Office/PDF conversion, rendering, OCR, arbitrary body-text editing, redaction, signing or certificate trust is claimed. All four editions are in_process local library execution; platforms stay self_hosted, linux, windows and mac, interfaces stay library, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims array was checked and left unchanged.

  324. d27989b

    The pdfcpu record, in the protect, merge and extract jobs, the owner and licence fields, the summary, the watchOutFor, the claim and the four editions

    What it said
    The owner field was empty although the README identifies Horst Rutter as maintainer. The summary and watchOutFor did not record the current v0.16.2 release, the Go 1.26 build requirement, the official binary matrix, the exact extraction limits or the separate signature-validation and removal capability. The unresolved convert job had no supporting prose or scope, and pdfcpu does not convert to other document formats. The protect and merge jobs had no scopes. The claim repeated the README opening sentence that pdfcpu supports signing, but the feature list and API expose only signature validation/evidence and removal. The four editions had unknown processing states.
    What it says now
    The owner is Horst Rutter and the licence is Apache-2.0. The current release is v0.16.2, published 11 October 2026; go.mod declares Go 1.26.0 and building requires Go 1.26 or later, with binaries built using Go 1.27.2. Official assets cover Darwin arm64/x86_64, Linux arm64/armv7/i386/x86_64, Windows i386/x86_64 and JS/WASM, plus source. The convert job is corrected to extract: pdfcpu documents image, font, content, page and metadata extraction with page selection, and does not extract plain text, perform OCR, offer attachment extraction or convert to other document formats. Protect covers encryption/decryption, owner/user password changes and permission list/set, including AES-256 handling described in the release notes. Merge covers create/append/zip modes with bookmarks, dividers, sorting and the separate split/trim/collect page assembly. Signature support is recorded separately as validation/evidence and removal through ValidateSignatures/RemoveSignatures, with a local trusted-certificate store and optional pdfcpu_eutl EUTL snapshot, no signing API and no legal/eIDAS/trust guarantee. The claim is corrected to signature validation/removal and marked partial. PDF 2.0 validation remains basic and improving; no general redaction workflow, rendering, OCR, generic conversion or arbitrary text editing is claimed. All four editions are in_process local CLI/library execution; platforms stay windows, mac, linux and self_hosted, interfaces stay library and cli, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true, openSource is true, selfHostable is true, dataResidency stays self_hosted and processesLocally is true.

  325. fc0e59f

    The PREP record, in the accessibility, OCR and convert jobs, the owner and licence fields, the price fields, the summary, the watchOutFor, the claims and the web and api editions

    What it said
    The owner was the brand Continual Engine rather than the legal entity Continual Engine US, LLC identified in the TX-RAMP certificate. The record did not record the TX-RAMP scope and dates, the 1 October 2025 Privacy/Fulfillment Policy dates, the API edition, the customer-specific on-premises bank deployment, or the Fulfillment Policy disclaimer of complete compliance. The accessibility, OCR and convert jobs had no scopes. The watchOutFor treated all uploads as leaving the environment and self-hosting as unestablished, while the on-premises case study shows one customer deployment; the API products were mentioned without an edition. The 100% and 60% claims already carried unverifiable verdicts but their findings needed the policy conflict; the cloud claim finding said nothing described on-premises, which the case study contradicts.
    What it says now
    The owner is Continual Engine US, LLC, with the brand Continual Engine, Austin/Texas footer address and Bengaluru office. The licence records the Privacy Policy and Fulfillment Policy/Terms of Service dates of 1 October 2025, the quote-only USD commercial model, no public DPA or security page, and no published retention or data-residency commitment. TX-RAMP Level 2 certification covers PREP and Invicta, granted 12 December 2025, expiring 11 December 2028, certificate TX1742040, and is not extended beyond that scope. The web and API editions are vendor_cloud with accounts and browser/http interfaces; platforms are web and api. Pricing is null/unknown and hasFreeTier is false because the vendor uses a quote/demo route. The accessibility job covers AI tagging, reading order, tables/lists/TOC/forms/math/alt text, accessibility checking/reporting and collaboration; OCR covers the built-in scanned-document OCR/correction/tagging workflow; convert covers the documented PDF-to-text/Word/HTML/ePub conversion and export paths. The 100% compliance and 60% turnaround claims are preserved verbatim with unverifiable verdicts, and the Fulfillment Policy disclaimer of complete compliance and legal advice is recorded. The cloud claim is confirmed with the customer-specific on-premises bank case study as a caveat; no public self-hosted edition is claimed, selfHostable stays null, dataResidency stays unknown and processesLocally stays false for the standard cloud service. No generic PDF editing, redaction, signing, universal ADA/WCAG/PDF/UA compliance or legal protection is claimed.

  326. 15e5c3f

    The PDF.js record, in the view and edit jobs, the owner and licence fields, the summary, the watchOutFor, the claims and the web and self_hosted editions

    What it said
    The owner field was empty although the project is the Mozilla PDF.js project. The summary and watchOutFor did not record the current v6.4.299 release, the npm pdfjs-dist distribution and its differing Node engine metadata, the exact annotation-editor types, the generic viewer visibility of signature/comment, the saveDocument path or the CORS/worker requirements. The view and edit jobs had no scopes, and the record did not distinguish the optional viewer from the core library or clarify that annotation edits can be saved into a PDF rather than being UI-only. The two editions had unknown processing states. The claims were current in substance but lacked the current editor/source detail.
    What it says now
    The owner is Mozilla PDF.js project and the licence is Apache-2.0, with pdfjs-dist recorded as the pre-built npm distribution rather than a hosted service. The current release is v6.4.299, published 3 October 2026; npm pdfjs-dist is 6.4.299. The npm package engine metadata is >=22.13.0 or >=24, while the repository package metadata is >=22.22.3, >=24.15.0 or >=26; the difference is recorded as a packaging/runtime distinction. The core library exposes getDocument, page.render and getTextContent, with same-origin/CORS restrictions for URL fetches and a worker requirement in browsers. The optional viewer adds navigation, search, print/download, form display/filling and annotation editing. The AnnotationEditorType enum includes FREETEXT, HIGHLIGHT, STAMP, INK, SIGNATURE, COMMENT and POPUP; underline and strikeout are rendered types rather than current editor types. The generic toolbar exposes free text, highlight, ink and stamp while signature and comment are hidden. The viewer save path uses PDFDocumentProxy.saveDocument and download, so annotations/forms can be written into a new PDF; there is no general high-level library editing API. The view job covers parsing/rendering/text extraction/viewer/search/print/download/form display; the edit job covers the viewer annotation editor and form filling with the exact types and save caveat. The signature editor is a visual annotation, not certificate-backed signing. No OCR, redaction, arbitrary body-text editing, merge/compress/encrypt, signing/trust or generic conversion is claimed. Both editions are in_process local/browser or Node execution; platforms stay web and self_hosted, interfaces stay library and browser, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier stays true, openSource stays true, selfHostable stays true, dataResidency stays self_hosted and processesLocally stays true. The claims were updated with the current annotation/source findings.

  327. eef8ff1

    The pdfplumber record, in the extract and view jobs, the owner and licence fields, the summary, the watchOutFor and the four editions

    What it said
    The owner field was empty although the LICENSE names Jeremy Singer-Vine. The summary and watchOutFor did not record the current 0.11.10 release, the exact pdfminer.six/Pillow/pypdfium2 dependencies, the PyPI Python floor versus the README tested range, the py3-none-any wheel, the CLI formats or the exact visual-debugging/read-only boundaries. The extract and view jobs had no scopes. The four editions had unknown processing states. The record did not explicitly state that there is no edit or convert job because upstream has no PDF generation, modification or OCR.
    What it says now
    The owner is Jeremy Singer-Vine and the licence is MIT, copyright 2015 Jeremy Singer-Vine, with pdfminer.six==20260107, Pillow>=12.2.0 and pypdfium2>=5.9.0 as direct dependencies. The current release is 0.11.10, published 15 June 2026. PyPI declares Python >=3.8 and a py3-none-any wheel, while the README says the library is tested on Python 3.10 through 3.14; the install floor and tested range are both recorded without treating them as contradictory. The extract job covers text/layout, words and characters with coordinates, tables, shapes, images, annotations/hyperlinks, metadata, password-protected PDFs and CLI CSV/JSON/text output, with no OCR, table-accuracy or scanned-document claim. The view job covers visual debugging through Page.to_image and debug_tablefinder plus the CLI text projection, not a general PDF reader or renderer. The catalog has no edit or convert job and none is added because upstream explicitly provides no PDF generation, modification or OCR and the CLI emits data rather than converted documents. All four editions are in_process local library/CLI execution; platforms stay windows, mac, linux and self_hosted, interfaces stay library and cli, pricingModel stays open_source, priceFromGbp stays null with priceBasis unknown, hasFreeTier is true, openSource is true, selfHostable is true, dataResidency stays self_hosted and processesLocally is true. The claims array was checked and left unchanged.

  328. 26710e6

    The Tabula record, in the extract and view jobs, the owner and licence fields, the summary, the watchOutFor, the interfaces and the four editions

    What it said
    The owner field was empty although the LICENSE names Manuel Aristaran. The summary and watchOutFor did not record the app release 1.2.1 versus the separately maintained engine 1.0.5 split, the engine 1.0.2 bundled in the app, the Java requirements, the bundled PDFBox/AGPL third-party licences, or the exact CSV/Excel boundary. The extract job had no scope, there was no view job for the local selection/preview workflow, the interfaces listed only CLI although the app uses a local graphical interface, and the public prose used the word web, creating a false platforms-versus-prose signal. The four editions had unknown processing states.
    What it says now
    The owner is Manuel Aristaran. The licence records MIT for the app and engine code and the bundled third-party boundary: the distributed app includes Apache PDFBox 1.8.0 under Apache-2.0, an AGPL-licensed component and other notices, while the engine pom references a newer PDFBox. The app release is 1.2.1, published 4 June 2018, and it bundles engine 1.0.2; the engine latest release is 1.0.5, published 17 August 2021. The app README describes a volunteer-run project unlikely to see updates; the engine receives occasional bug-fix releases. Windows and Linux need a Java Runtime Environment, the Mac app bundles Java, the 2018 distribution README says Java 6/7 compatible, and the engine source/target is Java 1.8. The extract job covers text-based table extraction and page/area, guess, columns, stream/lattice, password, batch and CSV/TSV/JSON CLI options, with no OCR, scanned-PDF, accuracy, edit or conversion claim. The view job covers the app local localhost selection and preview workflow, not a general reader or renderer. Interfaces are CLI and browser; platforms stay windows, mac, linux and self_hosted, with a local localhost description replacing the false web-platform prose. The site markets CSV and Excel, but the documented CLI formats are CSV/TSV/JSON and the app exports CSV, so no native XLSX writer is claimed. All four editions are in_process local execution; pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true, openSource is true, selfHostable is true, dataResidency stays self_hosted and processesLocally is true. The claims array was checked and left unchanged.

  329. 8d17618

    The PDFium record, in the view and edit jobs, the owner and licence fields, the summary, the watchOutFor and the four editions

    What it said
    The owner field was empty although the LICENSE copyright is The PDFium Authors. The record did not state that GitHub has no PDFium releases or use the Chromium DEPS pdfium_revision as the meaningful source snapshot. The summary and watchOutFor did not record the full BSD-3-Clause plus Apache-2.0/third-party licence text, the lack of official standalone binaries, the exact public edit/save/signature APIs, or the distinction between PDFium and Chromium wrappers/UI. The view job had no scope, the public prose used the word browser and created a false platforms-versus-prose candidate, and the four editions had unknown processing states. The edit scope existed but did not record the save flags, signature-inspection-only boundary or absence of OCR/redaction/encryption creation.
    What it says now
    The owner is The PDFium Authors and the licence is BSD-3-Clause plus the Apache-2.0 and third-party text in LICENSE, not BSD-only. GitHub has no PDFium releases; the record uses Chromium DEPS pin a50f1a5953264de698f398fe3a2ef6153677fafa as a moving source-revision snapshot and states that no standalone product version exists. The view job covers document loading, rendering, text extraction/search, forms, annotations and accessibility through the public API and Chromium wrappers, and is explicitly not a standalone viewer or OCR engine. The edit job covers FPDF_CreateNewDocument, FPDFPage_New, page-object creation/insertion and FPDFText_SetText, transforms/flattening, form filling, annotation creation and FPDF_SaveAsCopy with incremental/no-incremental/remove-security flags; it is a low-level embedding API, not a general Acrobat-style editor, and makes no arbitrary body-text layout editing, redaction, encryption-creation, signing or certificate-trust claim. Signature APIs are inspection-only, encrypted documents can have security removed but not encrypted, and OCR/Office conversion are not PDFium capabilities. Chromium wrappers and the Chrome viewer are distinguished from the engine; the test program is not an end-user application. No official standalone binaries are claimed; build support is Windows x64/x86, Linux x64, macOS and Android ARM with Chromium GN/Ninja/depot_tools. The browser word was removed from public prose to clear the false platform candidate. All four editions are in_process source/library execution; platforms stay windows, mac, linux and android, interfaces stay library, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true, openSource is true, selfHostable is true, dataResidency stays self_hosted and processesLocally is true. The claims array was checked and left unchanged.

  330. 2ac4271

    The SetaPDF-Core record, in the protect and extract jobs, the licence and price fields, the summary, the watchOutFor, the claims and the self_hosted edition

    What it said
    The price field carried GBP 154.4 as a synthetic conversion of the EUR 180 server licence. The record did not record the current PHP 7.2 minimum as of the 2.53.0 line, testing through PHP 8.6, the required extensions, the authenticated Composer repository and package name, the 08/2023 commercial licence terms or the precise server/cluster pricing and 20% update renewal. The protect and extract jobs had no scopes, and the self_hosted edition had an unknown processing state.
    What it says now
    The owner is Setasign GmbH Co. KG, Helmstedt, Germany. The licence is the commercial SetaPDF Product Range Software License Agreement dated 08/2023, governed by German law with Braunschweig jurisdiction and a German-original-prevails clause; it grants a perpetual server or cluster/cloud licence tied to identified servers, includes one year of updates and renews update access at 20%. Licensed users get source code access and do not need Ioncube; trials are Ioncube-encoded. Public pricing is EUR 180 for the server licence and EUR 270-810 for cluster/cloud by node count, exclusive of German VAT; no GBP conversion is made. Composer uses the authenticated Setasign repository and package setasign/setapdf-core (2.*), not Packagist. The current manual states PHP 7.2 minimum as of the 2.53.0 line and testing through PHP 8.6, with DOM, iconv or Multibyte String, OpenSSL, SPL, Zlib and intl. The public pages do not expose a more precise current build number, so the record keeps the 2.53 line and does not invent a revision. Protect covers Standard and Public-Key encryption/decryption, RC4-40/128, AES-128/256, user/owner passwords, permission flags, authentication and creation/removal of security handlers, with no signing or certificate-trust validation. Extract covers low-level object/metadata/page/canvas/annotation/outline/embedded-file inspection and reading, with no high-level text extraction, OCR, Office conversion or Acrobat-style UI. The two jobs are protect and extract; no edit job is added. The only edition is self_hosted in_process; interfaces stay library, platforms stay self_hosted, openSource is false, selfHostable is true, dataResidency stays self_hosted and processesLocally is true with the HTTP/stream-writer caveat. The current release is kept as the 2.53 line because no more precise public build was available. The claims were rechecked for low-level access and encryption.

  331. 2066b13

    The PDF Arranger record, in the merge and edit jobs, the owner and licence fields, the summary, the watchOutFor, the source string and the three editions

    What it said
    The owner field was empty although AppStream names The PDF Arranger team and setup.py names Jerome Robert as author. The licence was GPL-3.0 although AppStream and setup.py say GPL-3.0-or-later. The record did not record the current 1.14.0 release, the native Python GTK/PyGObject/Poppler/pikepdf stack, the Windows installer/portable/nightly and Linux/Flatpak/Snap/macOS packaging boundaries, or the documented manual features. The merge and edit jobs had no scopes, the interfaces field was empty without explaining the schema limitation, and the public prose created a false web/browser platform candidate. The three editions had unknown processing states.
    What it says now
    The owner is The PDF Arranger team, with Jerome Robert recorded as setup.py author. The licence is GPL-3.0-or-later per AppStream and setup.py, while GitHub metadata reports GPL-3.0 and the discrepancy is noted. The current public release is 1.14.0, published 23 May 2026; setup.py and the User Manual both state 1.14.0. The app is a Python GTK desktop application with native pikepdf>=6/libqpdf, GTK3, PyGObject/Cairo, Poppler GLib, libhandy, python-dateutil, packaging and optional img2pdf; it is not pure Python and is not hosted. Windows has official installer/portable ZIP and authenticated nightly builds; the Win32.md MSYS2 instructions are build prerequisites, not a universal support claim. Linux/BSD packages, Flatpak and Snap are listed; macOS uses Nix or manual Homebrew/GTK3 with no official signed package. Merge covers opening several PDFs, page thumbnails, reorder, copy/paste, reverse/swap, moving pages between instances, split/merge, booklets, selection and save/export. Edit covers whole-page rotation, deletion, duplication, blank pages, crop/margins, resize, metadata, overlay/underlay, image import, literal text/image clipboard extraction and image explosion, while excluding body-text editing, semantic extraction, OCR, redaction, signing, encryption, Office conversion, repair and hosted processing. Interfaces stay empty because the schema has no desktop/GUI value and the app is not a library, CLI, HTTP or hosted product; the false web/browser platform candidate is removed. All three editions are in_process local execution; platforms stay windows/mac/linux, selfHostable stays false, dataResidency stays self_hosted, processesLocally is true, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true and openSource is true. The claims array was already empty and remains empty; no claim was removed.

  332. 2ef050e

    The PAC record, in the accessibility and view jobs, the license field, the summary, the watchOutFor and the Windows edition

    What it said
    The view job had no scope even though the current product page documents opening, visual review, screen-reader/structure preview and summary/detailed reports. The accessibility scope was a 48-character summary rather than a source-backed account of the checks, reports and human-review limits. The platform candidate was triggered by web/browser/online wording in public prose, although PAC is Windows-only and the axes4 online/API/batch products are separate. The licence field was empty despite the FAQ free-of-charge statement and the General Terms, and the exact 2026.1/26.1.0.0 release and Windows 10/11/.NET 4.8 requirements were not carried in the summary/watchOutFor.
    What it says now
    The accessibility job covers machine-verifiable PDF/UA via the Matterhorn Protocol, WCAG 2.2 AA, quality checks, AI-assisted semantic checks, visual/manual review, screen-reader/structure preview and summary/detailed reports, while explicitly not repairing or editing and not proving legal accessibility. The view job covers opening a PDF, rendered visual review, screen-reader/structure preview and report display only, not general reading/editing, OCR, conversion, signing or repair. Public prose calls axesCheck a separate hosted tool and describes axesSense as a separate batch product, without web/browser/online wording that would imply PAC is a website; platforms stays [windows] and interfaces stays empty because PAC has no CLI or API. The licence is recorded as free of charge for all applications under the axes4 General Terms, with no public source distribution and separate axesCheck/axesSense/axesPAC products. The current release is 2026.1, build 26.1.0.0, dated 25 March 2026, for Windows 10/11 with .NET Framework 4.8 or higher. The FAQ offline/no-internet statement remains vendor-stated and not independently network-probed; no exact control count is repeated; a green report is not legal proof and human review remains required. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred. The two catalog jobs are accessibility and view; no edit, convert, OCR, sign or protect job is claimed. pricingModel stays free, priceFromGbp is null with priceBasis unknown, hasFreeTier is true, openSource is left unset, selfHostable is left unset, dataResidency stays self_hosted, processesLocally is true, interfaces stays empty and platforms stays windows.

  333. 3d5d288

    The Mindee record, in the extract and OCR jobs, the owner and licence fields, the price fields, the summary, the watchOutFor, the claims and the API edition

    What it said
    The owner field was empty although the legal terms name Mindee SAS for Eurozone clients and Mindee Inc. for others. The record did not establish V2 as the current line or distinguish V1 material, and did not carry the current EUR plans of Starter 44 and Pro 116, the 200-credit/14-day trial, the 10-page trial limit, the corrected no-Business-plan and no-old-179/584 position, or the current processing-zone, retention and deletion terms. The price field carried GBP 44 per month although no GBP price was verified, and hasFreeTier contradicted the internal trial note. The extract and OCR jobs had no scopes. The old verified GDPR/SOC 2 claim had a changed recheck, and the old homepage localized-zone/no-store claim is no longer present. The API edition had no processing note.
    What it says now
    The owner is Mindee SAS, with the licence field recording that Eurozone clients contract with Mindee SAS and all other clients with Mindee Inc. V2 is the latest documented line, with older V1 material still on platform.mindee.com; no public software version is published and none is invented. Current docs list Starter at EUR 44/month and Pro at EUR 116/month with annual billing, custom Enterprise, and a 200-credit or 14-day trial capped at 10 pages per document; hasFreeTier is false and no GBP price or conversion is asserted, so priceFromGbp is null and priceBasis unknown. The API is hosted asynchronous REST requiring an account and API key, with polling or webhooks and vendor-cloud processing; platforms stay api, interfaces http, processing vendor_cloud, requiresAccount yes, processesLocally false, selfHostable false and dataResidency multi. Processing zones may be no preference (default, may route to Europe or the United States), Europe or United States, are not available on all plans and cover processing only, not support or extraction/analysis; storage retention defaults to 12 hours with a 1-24 hour range, and files sent by API are not saved to disk. These are vendor statements and are not independently probed. Extract is scoped to ready-made/custom structured extraction, JSON fields, confidence, polygons, optional raw text and utility models; OCR is scoped to the dedicated Raw Text OCR model returning full-page text and per-word positions via the async API, with no layout/accuracy guarantee and no structured-extraction conflation. The old verified GDPR/SOC 2 claim is retained only as a reported qualified vendor statement; the old homepage localized-zone/no-store claim was removed because it is no longer present. The separate open-source docTR library is not the API and does not make Mindee self-hostable. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  334. 5fd004b

    The Nougat record, in the extract and OCR jobs, the owner and licence fields, the summary, the watchOutFor, the source string and the self_hosted edition

    What it said
    The owner was the bare Meta label rather than the formally supported Meta Platforms, Inc. attribution. The record did not distinguish the PyPI package version from the repository source version and the GitHub model-weight releases, and did not record the package/source/model-tag split, the local nougat_api endpoint with start/stop, the Python/PyTorch/transformers/pypdfium2 runtime boundary, or the exact CLI/page/model-tag and [MISSING_PAGE]/--no-skipping behaviour. The extract and OCR jobs had no scopes, and the self_hosted edition had an unknown processing state.
    What it says now
    The owner is Meta Platforms, Inc., supported by the repository LICENSE copyright and project page. The package is nougat-ocr; PyPI currently serves 0.1.17 (uploaded 2023-10-04), the repository source declares 0.1.18, and GitHub releases 0.1.0-small and 0.1.0-base are model weights published 22 August 2023, not a software release. The README states the code is MIT and the weights are CC-BY-NC, with no commercial weights option. Extract covers academic PDF transcription to Markdown/Mathpix-compatible .mmd with LaTeX maths and tables, CLI input from files, directories or line-delimited lists, page ranges for single PDFs, model tag selection and [MISSING_PAGE]/--no-skipping caveats. OCR covers neural academic-document transcription, raw page text/equations/tables, and the optional local nougat_api server at 127.0.0.1:8503/predict with start/stop, with English best, other Latin-based languages possibly working, Chinese/Russian/Japanese not supported, and no general image OCR or broad-language claim. Weights may download on first use and practical use generally needs a GPU; Python >=3.7 and PyTorch/transformers/pypdfium2 dependencies are loaded from setup metadata. Platforms stay self_hosted, interfaces stay library/cli/http, selfHostable is true, processesLocally is true, dataResidency stays self_hosted, openSource is true for the code while the weights remain non-commercial, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true and launchedYear stays 2023. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred. The two claims were checked against the current README and remain supported; none was removed.

  335. faf3a58

    The pdftk-java record, in the protect and edit jobs, the owner and licence fields, the summary, the watchOutFor, the bestFor, the source and the four editions

    What it said
    The owner was empty although the source headers identify Marc Vinyals as the Java-port author and Steward and Lee, LLC as the original PDFtk copyright holder. The record did not distinguish pdftk-java tag/program version 3.3.3 from the original PDFtk attribution, did not carry the GPL-2.0-or-later plus bundled iText/Batik/BouncyCastle third-party boundary or the original PDFtk commercial-redistribution note, called the Java build maintained despite the newest release being 2022, and did not record the CLI-only/non-library guidance or the Java 8/JRE packaging. The protect and edit jobs had no scopes, and the four editions had unknown processing states. The seamless-replacement claim lacked the full FAQ caveat.
    What it says now
    The owner is Marc Vinyals, with the original PDFtk/Steward and Lee attribution and bundled third-party licences kept separate. The newest tag/release is v3.3.3 dated 22 September 2022 and no newer release exists; the source/program version is 3.3.3, printed as pdftk port to java 3.3.3, with no separate original-PDFtk version printed. The code is GPL-2.0-or-later; the bundled modified iText, Apache Batik, BouncyCastle and other components retain their own terms, and the original PDFtk licence note requires a commercial licence for commercial redistribution as part of another product unless that product is GPL-compatible, while ordinary local use remains free. The README positions pdftk-java as a command-line tool and recommends iText or OpenPDF for a Java library API. Protect covers 40-bit/128-bit RC4, AES-128 default with passwords, owner/user passwords, permission flags and decryption, with no signature, DRM or redaction claim. Edit covers rotation, FDF/XFDF form fill/flatten, background/multibackground, stamp/multistamp, metadata/bookmark/page-data updates, attachments/unpacking and compress/uncompress stream filters, with page assembly/cat/shuffle/burst left in the untouched merge scope and no body-text/OCR/Office/repair/signing claim. The seamless-replacement FAQ claim is retained with the CLI and non-library caveat. Platforms stay windows/mac/linux/self_hosted, interfaces stay cli, selfHostable is true, processesLocally is true, dataResidency stays self_hosted, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown, hasFreeTier is true and launchedYear stays 2018. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  336. 3a575ea

    The Xournal++ record, in the view and convert jobs, the owner and licence fields, the summary, the watchOutFor, the source string and the Windows/macOS/Linux editions

    What it said
    The owner was empty although the AUTHORS file credits the Xournal++ developer team and community. The record did not carry the current 1.3.8 release, relied on a stale generated 1.3.7 man page for CLI evidence, and did not scope or name the view and convert jobs. The licence was GPL-2.0 but lacked the community-project attribution. The PDF viewing/navigation/outline and export-flattening caveats were only partly recorded, and the .xopp absolute-path/attach-mode and separate Xournal++ Mobile boundaries were not fully reconciled. The three editions had unknown processing states.
    What it says now
    The owner is Xournal++ contributors, a community project, with GPL-2.0 recorded from LICENSE and GitHub metadata. The current release is v1.3.8, published 26 September 2026; the changelog also lists 1.3.7 and 1.3.6. The repository no longer has a checked-in current 1.3.8 man page; current README/source documents --create-pdf, --create-img and PNG sizing options, and the old generated 1.3.7 page is stale. View covers loading a PDF as a journal background, page viewing/navigation, links, background-text search/selection/copy and outline/structure inspection, with no general PDF editing, OCR, semantic extraction or PDF-native annotation preservation. Convert covers flattened/rendered export to PDF, PNG and SVG from the GUI and command line, with documented caveats that annotations become non-editable, PDF-native annotations/forms are not preserved, background/paper-style settings can change output, and no Office/DOCX/PDF-to-text/OCR/general conversion is claimed. Journals are .xopp files referencing the background PDF by absolute path unless attach mode is used; Xournal++ Mobile is a separate project and is not mixed into this desktop record or public prose. The existing edit scope is preserved. Platforms stay windows/mac/linux, interfaces stay cli, selfHostable remains unset, dataResidency stays self_hosted, processesLocally is true, openSource is true, pricingModel stays open_source, priceFromGbp is null with priceBasis unknown and hasFreeTier is true. The claims were checked against current sources and left unchanged; no claim was removed. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  337. ae2f1cc

    The pdfToolbox record, in the accessibility and edit jobs, the editions, platforms, interfaces, processing/locality fields, summary, watchOutFor and source

    What it said
    The record had a remote vendor_cloud/http edition described as Server and CLI workflows with activation and update checks, which conflated licence/update checks with document processing and contradicted the self-hosted Desktop/Server/CLI/SDK product. The local edition claimed Linux in a Desktop-named edition although the sources place Linux under Server/CLI/SDK, Desktop was not limited to Windows/macOS, the SDK and Server/CLI packaging were described as four unrelated purchases, and the optional OEM cloud API was not separated from the normal product. hasFreeTier was unset despite only a 14-day trial, current version 17.1.692 was not recorded, and the accessibility and edit jobs had no scopes.
    What it says now
    Desktop is a local in_process manual preflight/correction application on Windows/macOS, standalone or an Acrobat Pro plug-in; activation/update checks do not make it vendor cloud. Server/CLI/SDK is a customer_server deployment on Windows/Linux/macOS/self_hosted with CLI, library and HTTP/REST/Docker interfaces, including Desktop for profile configuration in the Server/CLI package and Server using CLI behind the scenes. pdfToolbox in the Cloud/OEM API is a separate OEM-only, AWS-managed service at api.callassoftware.io with x-api-key and requires an account; it is vendor_cloud/api/http and is not the normal product. Platforms are windows/mac/linux/api/self_hosted and interfaces cli/library/http. processesLocally is true for Desktop/Server/CLI/SDK with the cloud caveat in the note; dataResidency stays self_hosted; product-level requiresAccount is false and the cloud edition requires an account. hasFreeTier is false because only the 14-day trial exists; pricing remains quote-based and no price is invented. The current release is 17.1.692, dated 1 October 2026. Accessibility is scoped to PDF/UA conformance checks/validation and profile-driven mechanical fixes, not semantic accessibility remediation or a screen-reader authoring editor. Edit covers profile-driven correction/fixup, process plans and documented validation/optimization/conversion, with manual Desktop correction distinct from automated Server/CLI/SDK processing and no generic body-text editing claim. The existing verified profile/standards claim was kept. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  338. 78578b5

    The PDFix record, in the accessibility and edit jobs, the owner and licence fields, the price/free fields, the summary, watchOutFor, platforms, interfaces, editions and claims

    What it said
    The owner was empty although the EULA names PDFix s.r.o. The record carried four platform/API editions with unknown processing, including a PDFix-hosted API row, although Desktop and SDK are local products and there is no hosted API edition. The price field carried GBP 299.53 as a synthetic EUR conversion, and the record did not record the current Desktop 3.3.0/SDK 9.3.0 versions, the Desktop Lite/SDK Lite free tiers, the watermarked free download, the separate 30-day Trial License, the current EUR/USD prices or the SDK-only redaction boundary. The locality claim was absolute despite the documented Amazon Textract/OpenAI external actions. The accessibility and edit jobs had no scopes. All three claims remained marked verified even though their 2026-09-22 rechecks said the wording had changed or was absent.
    What it says now
    The owner is PDFix s.r.o., Jilemnick�ho 2346/3, 90031 Stupava, Slovak Republic, CORP ID 46248650, with PDFix-US LLC in Frederick, Maryland as the US distributor. Current official downloads identify PDFix Desktop 3.3.0 and SDK 9.3.0 on Windows/macOS/Linux. Desktop is a local GUI accessibility remediation/checking application and SDK is a local embedded library with CLI and Python, C#, Java, C++, Node.js interfaces; the api edition/platform was removed because there is no PDFix-hosted REST/SaaS product. Core processing is local/on-premises, with the explicit exception that optional AI Actions can send page images/content to Amazon Textract or OpenAI; the locality note records both. Desktop Lite and SDK Lite are permanently free, free downloads can be used without registration or time limit but saved documents can be watermarked, and the EULA separately defines a 30-day Trial License. Desktop Pro is EUR 350 or USD 599 per user per year; Enterprise is EUR 1,000 per user per year; SDK tiers are quote-only. priceFromGbp is null with priceBasis unknown, hasFreeTier is true, and no synthetic GBP conversion is made. Redaction is SDK Professional/Enterprise only. The vendor does not guarantee 100% accessibility; validation and mechanical fixes require human semantic review. Accessibility covers PDF/UA, WCAG 2.1/2.2, Section 508/EAA, auto-tagging, manual remediation, reading order, alt text, headings, tables/lists, forms and metadata; edit covers tagging/structure/content remediation and SDK actions, not generic body-text editing. All three old verified claims are downgraded to reported with current qualifiers, including the core-vs-external-AI distinction, free-download-vs-Lite-vs-trial distinction and SDK Lite scope. platforms are windows/mac/linux, interfaces library/cli, selfHostable true, processesLocally true with the external-action exception, dataResidency self_hosted, requiresAccount false, openSource null, pricingModel paid and launchedYear remains unknown. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  339. ecca039

    The pdfminer.six record, in the extract job, the owner, summary, watchOutFor, platforms, interfaces, processing notes, source and four editions

    What it said
    The owner was empty although the project is a community-maintained fork with no company owner. The record did not carry the current PyPI 20260107 release, Python >=3.10 requirement, py3-none-any wheel, charset-normalizer dependency or the optional Pillow image extra. The extract job had no scope. The localities note made an absolute no-network claim, the four editions had unknown processing and account status, and the interfaces were only at record level. The summary did not distinguish the stable package from the newer documentation development version.
    What it says now
    The owner is pdfminer.six contributors, a community-maintained project, with MIT and the 2004-2016 Yusuke Shinyama copyright recorded. The current PyPI release is 20260107, uploaded 7 January 2026, requiring Python >=3.10 and publishing a py3-none-any wheel; installation brings charset-normalizer and cryptography, and Pillow only under the optional image extra. The project code is pure Python but not dependency-free. The extract scope records text with location/font/colour, embedded images, HTML/hOCR, automatic layout analysis, PDF object parsing, AcroForm form extraction, table-of-contents/outline, tagged content, CJK/vertical writing, PDF-1.7 almost-complete support, RC4/AES input, and pdf2txt.py/dumppdf.py. It explicitly excludes OCR, PDF writing/editing, redaction, signing, repair and semantic/reading-order guarantees; image-only scans need separate OCR, tables/columns can interleave, and (cid:x) values are font/Unicode-mapping gaps. All four editions are in_process with library/cli interfaces and no account; the locality note says processing runs in the local Python interpreter with no PDFminer service/account, avoiding an absolute no-network guarantee. Platforms stay linux/windows/mac/self_hosted. The two existing claims were checked against the current README and remain current. launchedYear remains null and lastReviewedAt/page-lastmod remain unknown because no publication date is inferred.

  340. e706426

    The PDFShift record, in the edition, platform, interface, processing, maxFileSizeMb, price, summary, watchOutFor and convert scope

    What it said
    The record contained a duplicate maxFileSizeMb: the canonical attributes value 250 and a stale top-level null. The top-level GBP 6.68 month price was a synthetic conversion with no verified GBP source. The edition had an empty interface list and no source, the public prose used browser wording that created a false platform candidate, and the convert scope was only a short website-capture line that did not distinguish HTML/raw HTML/URL PDF conversion from the documented image/screenshot output modes. The summary and watchOutFor carried an unresolved no-file-size-limit speculation that the current pricing endpoint resolves to 250 MB paid/15 MB free.
    What it says now
    The duplicate is resolved: attributes.maxFileSizeMb is 250 MB and the top-level duplicate is gone. The free plan is 50 credits/month, 15 MB and 30 s; paid plans currently have 250 MB and 100 s, with Startup USD 9/500 credits and larger tiers from the current pricing endpoint. priceFromGbp is null, priceBasis unknown and hasFreeTier true, with no synthetic GBP conversion. The API is a single vendor_cloud edition at api.pdfshift.io with http interface, account/X-API-Key, and no self-hosted or local edition. Default processing is in memory/no traces, with documented filename/webhook S3 storage, log_request raw-request storage and s3_destination/is_gdpr/is_hipaa exceptions; EU-customer OVH France storage is stated, but non-EU residency/S3 region and broader transfer caveats are explicit. Authenticated free conversions are not watermarked; sandbox/unauthenticated requests can be. No SLA is guaranteed. Owner PDFShift SASU and launchedYear 2018 are supported by current terms/structured data. Convert covers HTML/raw HTML/URL to PDF at /v3/convert/pdf plus documented screenshot/PNG/JPEG/WEBP/Open Graph output modes where supported, credit accounting by generated output size, and explicitly excludes Office/Word/Excel conversion, generic PDF editing, OCR, redaction and signing. The two existing claims were checked and left unchanged. lastReviewedAt and page-lastmod remain unknown because no publication date is inferred.

  341. 8b080f0

    The PDFShift dataResidency field and processing note

    What it said
    dataResidency was eu after the processing note was corrected to say non-EU residency and the S3 storage region were not established, so the record asserted an EU-only residency while its own evidence note doubted it.
    What it says now
    dataResidency is unknown. The vendor statement that EU customer data is stored in OVH France is preserved in the processing note and watchOutFor, along with the explicit facts that non-EU customer residency and the S3 region are not established and the privacy terms contain broader transfer caveats. The record no longer implies all customers are physically in France.

  342. 30d7ba9

    The Prince record, in the convert job, price/free fields, summary, watchOutFor, platforms, interfaces, installed edition, processing/locality and claims

    What it said
    The record carried a synthetic GBP 370.42 one-off conversion of the USD 495 Desktop price, while internalNotes said the price was null and the conversion was only a snapshot. It had a placeholder self_hosted edition with unknown processing and no interfaces/source, listed only self_hosted platforms, did not record Prince 17/17-1, and its public prose used browser wording that created a false platform candidate. The convert job had no scope. Desktop USD 495 was not distinguished from server or commercial-site licensing, and the narrow Non-commercial licence restrictions were not represented in a claim.
    What it says now
    The record is for installed Prince 17 (17-1 packages; no exact patch date exposed). The one installed edition has CLI/library interfaces, in_process processing and Windows/macOS/Linux/self_hosted platforms; FreeBSD and AWS Lambda are noted as download/deployment targets, not invented schema platforms, and no Prince-hosted API edition exists. requiresAccount is false because local use needs no account, though a licence file may be required. priceFromGbp is null and priceBasis unknown; hasFreeTier is true only for the restricted Non-commercial licence. Current USD tiers are Desktop 495 one-time for one user/one computer self-publishing/internal use, Per Server 3,800 one-time for government/noncommercial internal/free documents, Academic Server 1,900, Commercial Site from about 2,000/year for startups, OEM quote-based, with no GBP conversion. The convert scope covers HTML/XML/CSS-to-PDF with Markdown/SVG/JS and JPEG/PNG/WebP/AVIF/GIF/TIFF inputs, PDF bookmarks/links/metadata/encryption/security/font embedding/attachments, CLI and wrapper libraries, and excludes generic PDF editing, OCR, redaction, signing, Office/Word conversion and hosted services. The existing HTML-to-PDF and Prince for Books claims remain current; a reported claim records the Non-commercial restrictions and notice/link obligations. Prince Cloud, DocRaptor and EuroPDF are separate services. launchedYear remains null and lastReviewedAt/page-lastmod remain unknown.

  343. 2cf7a91

    The Prince watchOutFor prose

    What it said
    The watchOutFor still contained the literal word browser even after the browser-fidelity wording was otherwise removed, so the platform candidate remained unresolved.
    What it says now
    The sentence now says Prince is its own typesetting engine rather than a general-purpose renderer, and that no fidelity comparison against a general-purpose renderer was run. No browser wording remains in public prose and platforms stay windows/mac/linux/self_hosted.

  344. c1bcbda

    The Prince watchOutFor prose

    What it said
    The prior prose-cleanup attempt did not actually write the file because its JavaScript contained a PowerShell-style test; the literal word browser therefore remained in watchOutFor and the platform candidate stayed unresolved. The prior ledger entry recorded the intended cleanup before it was actually applied.
    What it says now
    The sentence now says Prince is its own typesetting engine rather than a general-purpose renderer, and that no fidelity comparison against a general-purpose renderer was run. No browser/web/online trigger remains in public prose, platforms stay windows/mac/linux/self_hosted, and the installed-only edition and convert scope are unchanged.

  345. d825046

    The ReportLab record, in the protect and accessibility jobs, version/runtime, summary, watchOutFor, editions, dataResidency, claims and pricing/licence split

    What it said
    The record had no current version, and the protect and accessibility jobs had no scopes. dataResidency was unknown for an in-process library. The first claim was a synthetic assembled quotation from a comparison table and was marked verified; the nag-line claim was verified without the evaluation-registration/licence-file conditions and could be read as a universal free-tier watermark. The edition read dates and licence components were stale, and the open-source/PLUS accessibility split was not scoped.
    What it says now
    The record carries version 5.0.1 (20 August 2026), Python >=3.9 and <4, BSD, py3-none-any, with accel/pycairo/bidi/shaping extras and base Pillow/charset-normalizer. The BSD toolkit and proprietary rlextra PLUS are separate in-process library editions; no hosted API or CLI is invented. dataResidency is self_hosted for those packages, with ReportLab separately sold hosted solutions excluded. protect covers StandardEncryption with user/owner passwords and print/modify/copy/annotate permissions on generated PDFs only, not redaction/content removal or arbitrary existing files. accessibility records that the open-source toolkit generates ordinary PDFs while auto-tagging/tagged=1/RML controls are PLUS 4+, and that tagging/validation is not semantic or legal remediation; no validator is claimed. priceFromGbp remains null for the BSD package; PLUS GBP 1,728/year remains separated in internal notes. The two claims are downgraded to reported: the toolkit claim is now a paraphrase from current docs, and the nag-line claim carries the evaluation-registration/licence-file conditions and is not a universal free-tier watermark. Owner remains null because current pages conflict between ReportLab Inc. and ReportLab Europe Ltd.; launchedYear stays 2000; qualityScore remains null. lastReviewedAt/page-lastmod remain unknown.

  346. abe5210

    The CommonLook PDF record, in the edit and accessibility jobs, version, claims, summary, watchOutFor, source, editions and interfaces

    What it said
    The edit job was an unresolved prose-only candidate because the record said editing was part of remediation without defining it. The version was absent despite the current resource hub publishing desktop version 5.17. Both 2026-09-22 claims were still marked verified: one was current vendor wording but had no fresh evidence file, and the other was unsupported marketing about tens of thousands of customers and standards leadership. Edition sources/read dates were stale, and the edit boundary between accessibility-structure remediation and generic content editing was not stated.
    What it says now
    The edit job is scoped to editing an existing PDF accessibility structure in the Advanced or Simplified editor: tag tree, reading order, alt text, headings, tables/lists and form-field labels, followed by revalidation. It is not generic body-text/word-processor authoring, OCR or arbitrary content editing; OCR and interactive-element creation are separate pre-remediation steps. The accessibility job remains scoped to AI/manual tagging, structural remediation and PDF/UA/WCAG/HHS validation with a downloadable report, with no semantic or legal guarantee. version is 5.17 for the desktop product (published 11 September 2024); the web app version is not published and is not asserted. Owner stays Allyant, with NetCentric Technologies Inc. d/b/a CommonLook as the legal entity. Pricing remains quote-only with priceFromGbp null, hasFreeTier false and a 7-day trial that is not a free tier; the separate CommonLook PDF Validator is not this product. The web edition is upload-based with US/Canada processing/storage; the desktop edition is an Acrobat plug-in, and no absolute no-upload guarantee is inferred. Both 2026-09-22 claims are downgraded to reported: the standards-support sentence remains current vendor wording but no fresh evidence file was archived, and the customer-count/standards-leadership sentence is unsupported marketing. qualityScore remains null with no score invented. launchedYear and lastReviewedAt/page-lastmod remain unknown.

  347. 7facbdb

    The Documenso record, in the edit job, version, summary, bestFor, watchOutFor, platforms, interfaces, processing, editions and claims

    What it said
    The edit job was an unresolved prose-only candidate because the record described a signing platform as not editing documents and still carried an edit membership signal. Version was absent despite release 2.20.0. The web/api/self-hosted editions had unknown processing and no sources, the product carried an unsupported library interface, and both 2026-09-22 claims were still verified: one unsupported marketing statement about being the world most trusted signing tool, and one current self-host deployment statement without a fresh evidence file.
    What it says now
    The edit job is removed. Documenso uploads PDFs as-is (or converts DOCX to PDF), then prepares recipients, fields, templates and workflow settings and collects signatures; current docs do not document body-text, image, layout, redaction or OCR editing. Summary/bestFor/watchOutFor state that it signs and tracks rather than modifying visible content and do not imply a general content or accessibility editor. Version 2.20.0 (8 October 2026) is recorded. Owner remains Documenso, Inc. per current terms/privacy. The editions are hosted web (vendor_cloud), hosted REST API (vendor_cloud, http) and self-hosted Community/Business/Enterprise (customer_server, http); the unsupported library interface is removed. requiresAccount is true; processesLocally is false for the shared default with self-hosted processing explained; selfHostable true; openSource true; dataResidency eu based on the specific GDPR page (EU application data, document storage and backups), with the generic privacy policy US transfer clause recorded as a conflict to check contractually. priceFromGbp is null and hasFreeTier true; hosted Free/Individual/Teams/Platform prices remain in the record but are not assigned to the AGPL edition. Both claims are downgraded to reported: the world-most-trusted sentence is unsupported marketing, and the self-host deployment sentence is current but has no freshly archived evidence file. qualityScore remains null; launchedYear stays 2023; lastReviewedAt/page-lastmod remain unknown.

  348. f4a0038

    The pdf-lib record, in the edit job, version, owner, summary, watchOutFor, platforms, interfaces, editions, processing/locality and maintenance facts

    What it said
    The primary edit job was an unresolved prose-only candidate. version was absent despite npm/GitHub release 1.17.1, owner was empty despite the npm author metadata, the web and self-hosted editions had unknown processing and no sources, and the public prose did not define the boundary between object-level PDF modification and body-text editing, OCR, redaction or a general editor.
    What it says now
    The edit job is scoped to loading and modifying an existing PDF with pdf-lib object-level operations: adding/inserting/removing/copying pages, drawing new text/images/SVG/embedded PDF pages, creating/filling/reading/flattening AcroForm fields, metadata/viewer preferences, attachments and font embedding/subsetting. It explicitly excludes reflowing or editing existing body text outside form fields, OCR, redaction, encryption/decryption, digital signatures, accessibility remediation and general word-processor editing. version is 1.17.1 (latest release 6 November 2021; repository last pushed 17 July 2024, 318 open issues). Owner is Andrew Dillon from npm metadata; no company or legal entity is documented. platforms are web/self_hosted and interfaces library/browser; both editions are in_process with no account or vendor service. dataResidency is self_hosted, processesLocally is true with a qualified caller-network note, requiresAccount false, openSource true, selfHostable true, priceFromGbp null, hasFreeTier true. The encryption and modification claims remain current; none was downgraded. qualityScore remains null; launchedYear stays 2017; lastReviewedAt/page-lastmod remain unknown.

  349. 6d7e1f9

    The pdf-redactor record, in the protect job, version, owner, summary, watchOutFor, editions, interfaces, processing/locality and maintenance facts

    What it said
    The primary protect job was unresolved because it had no scope and no protect/redaction word in finder prose. version was absent despite PyPI 0.0.1, owner was empty despite the package author metadata, and the three OS editions had unknown processing and no sources. The record did not define protect as text/annotation/link/metadata redaction rather than encryption or complete sanitisation.
    What it says now
    protect is scoped to documented text-layer redaction via regex/filter functions, plain-text annotations, link target URLs, the Document Information Dictionary and embedded XMP metadata, with metadata rewrite/removal/addition and a new PDF written. It explicitly excludes password encryption, permissions, complete sanitisation, image/graphic sanitisation, embedded files/multimedia/scripts/rich-text annotations/forms/internal object names/signatures, and preserves qpdf/font caveats and the no-op CLI limitation. version is 0.0.1 (19 April 2021), Python >=3.7,<4, CC0-1.0, deps pdfrw/defusedxml. Owner is Joshua Tauberer, with Antoine McGrath credited as co-author; no company is documented. platforms are windows/mac/linux and interfaces library; all editions are in_process, no account, no hosted service. dataResidency is self_hosted, processesLocally is true with a qualified caller-network note, requiresAccount false, openSource true, selfHostable true, priceFromGbp null, hasFreeTier true. The three existing claims remain current and verified; none was downgraded. No convert/edit/extract/OCR/accessibility/signing job is invented. qualityScore remains null; launchedYear stays 2016; lastReviewedAt/page-lastmod remain unknown.

  350. 9c4c3ac

    The OCRmyPDF record, in the OCR job, version, owner, summary, watchOutFor, platforms, interfaces, editions, processing/locality and dependency facts

    What it said
    The primary OCR job was an unresolved prose-only candidate. version was absent despite PyPI/GitHub release 17.13.0, owner was empty despite the package author metadata, and the four platform editions had unknown processing and no sources. The record did not define OCR as searchable text-layer generation rather than generic text extraction, body editing, redaction or semantic accessibility.
    What it says now
    The OCR job is scoped to adding a searchable OCR text layer to existing scanned PDFs with OCRmyPDF: rasterising pages, running Tesseract with requested language packs, and writing recognised text as an invisible layer, with documented deskew, page rotation, unpaper cleaning, image optimisation, PDF/A output, sidecar OCR text and batch/hot-folder workflows. It explicitly excludes generic text extraction to arbitrary formats, editing/reflowing visible body text, handwriting OCR, redaction/content removal, digital signatures, semantic accessibility remediation/validation and arbitrary scanned-content understanding. version is 17.13.0 (28 September 2026), Python >=3.11, MPL-2.0, author James R. Barlow. platforms are windows/mac/linux/self_hosted and interfaces library/cli; all editions are in_process, no account, no vendor service. dataResidency is self_hosted, processesLocally is true with a qualified caller-network note, requiresAccount false, openSource true, selfHostable true, priceFromGbp null, hasFreeTier true. The bundled Flask/Streamlit wrapper is documented as a demo, not a hosted service. The record had no claims entries; none was invented. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  351. b278aa4

    The OCRmyPDF watchOutFor prose

    What it said
    The watchOutFor said the bundled Flask/Streamlit web wrapper, which used the word web in public prose while platforms are windows/mac/linux/self_hosted; the sweep treated that as a platform-versus-prose contradiction.
    What it says now
    The sentence says the bundled Flask/Streamlit HTTP demo wrapper has no authentication and is not for public deployment. No web/browser/online trigger remains in public prose, and platforms stay windows/mac/linux/self_hosted because OCRmyPDF is not a hosted web product.

  352. 4302a82

    The mPDF record, in the platforms-vs-prose wording, version, summary, watchOutFor, source, licence and self-hosted edition

    What it said
    The record correctly listed platforms as self_hosted, but public prose used web server, browser print dialog, headless Chrome and browser-level sanitisation language, which made the finder look for a website platform that does not exist. version was absent despite Packagist 8.3.1, PHP requirements were not recorded, and the self-hosted edition had unknown processing and no source.
    What it says now
    mPDF remains a self-hosted PHP library only. The prose now uses PHP server process, general-purpose renderer, headless Chromium renderer and ordinary HTML escaping language, so no web/browser platform is implied and no web edition is added. version is 8.3.1 (Packagist, 11 March 2026), GPL-2.0-only, requiring PHP ^5.6/^7/~8.0-8.5, ext-gd and ext-mbstring. The self_hosted edition is in_process, library-only, no account, no vendor service; processesLocally is true with a caller-network qualification, dataResidency is self_hosted, interfaces library, priceFromGbp null, hasFreeTier true, openSource true, selfHostable true. Owner remains null because the project is community-maintained and the README credits Ian Back, with no company documented. The two existing claims remain current and verified. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  353. 8ba66b8

    The WeasyPrint record, in the platforms-vs-prose wording, version, requirements, summary, watchOutFor, source, convert scope and claims

    What it said
    The record correctly listed platforms as self_hosted, but public prose used web developers and browser language, which made the finder look for a website platform that does not exist. version was absent despite release 70.0, Python/Pango requirements were not recorded, the convert job had no scope, and both claims were still verified even though one was marketing and the other an unsupported 15-years figure.
    What it says now
    WeasyPrint remains a self-hosted Python library/CLI only. The prose now uses HTML/CSS, general-purpose renderer and renderer-specific language, so no web/browser platform is implied and no web edition is added. version is 70.0 (8 September 2026), Python >=3.10, Pango >=1.44, BSD-3-Clause, with pydyf/cffi/tinyhtml5/tinycss2/cssselect2/Pyphen/Pillow/fonttools dependencies. The self_hosted edition is in_process, library/cli, no account, no vendor service; processesLocally true with a caller-network qualification, dataResidency self_hosted, interfaces library/cli, priceFromGbp null, hasFreeTier true, openSource true, selfHostable true. The convert job is scoped to HTML/CSS-to-PDF generation, PDF/A and PDF/UA variants, tagged output, forms/attachments/metadata and CLI options, with explicit no-JavaScript/browser-engine, no existing-PDF editing, no extraction/OCR/redaction/signature and no legal-accessibility-certification boundaries. Both claims are downgraded to reported: the smart-solution sentence is marketing and the 15-years figure is unsupported. qualityScore remains null; launchedYear stays 2011; lastReviewedAt/page-lastmod remain unknown.

  354. 6588842

    The WeasyPrint summary and watchOutFor prose

    What it said
    The first cleanup still used browser engine and not a browser in the summary and web edition in watchOutFor, so the platforms-vs-prose candidate remained unresolved despite platforms correctly being self_hosted.
    What it says now
    The summary says a full general-purpose rendering engine and not a general-purpose renderer; watchOutFor says no hosted edition. No web/browser/online trigger remains in public prose, and platforms stay self_hosted because WeasyPrint is a self-hosted Python library/CLI, not a website product.

  355. 72dd753

    The pdfGoHTML record, in the accessibility job, platforms-vs-prose wording, version, summary, watchOutFor, editions, interfaces and claim

    What it said
    The accessibility job was an unresolved prose-only candidate even though the product is only a human inspection aid, and the platforms-vs-prose candidate came from browser wording about the exported HTML views rather than a real web edition. version was absent despite release 2.2.032, and the editions had unknown processing and no sources. The single claim was verified without a fresh evidence file.
    What it says now
    The accessibility job is scoped to human inspection of tagged PDFs via structured HTML views - default, structure tags, easy-reader, inverted, low-vision and dyslexia-friendly - covering tag structure, language, content order and archival structure review. It explicitly does not validate PDF/UA or WCAG, remediate/edit tags, certify accessibility, replace screen-reader testing or guarantee legal compliance. platforms remain windows/mac; the prose now says default view rather than browser view, so no web edition is implied. version is 2.2.032 (4 September 2025); owner remains callas software GmbH. interfaces is empty (GUI Acrobat plug-in); both editions are in_process, account is for the download only, no hosted service. dataResidency self_hosted; processesLocally true with no transmission claim; requiresAccount true; openSource false; selfHostable true; pricingModel free; priceFromGbp null; hasFreeTier true. The claim is downgraded to reported because no fresh evidence file was archived, while recording that the wording is current. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  356. 9a49bcb

    The OpenSign record, in the sign job, version, licence, summary, watchOutFor, editions, interfaces, processing/locality, pricing and claims

    What it said
    The sign job was an unresolved prose-only candidate. version was absent despite release v2.41.3, the hosted/API/self-hosted editions had no sources beyond placeholder notes, interfaces were only at record level, and the licence was stated as simple AGPL-3 without the repository carve-outs. Pricing/free-tier and locality distinctions were partly unverified.
    What it says now
    The sign job is scoped to the documented server-backed e-signature workflow: upload/prepare a PDF, recipients, signing fields/templates, signing order/links, typed/drawn/uploaded signatures, emailed one-time-code guest verification, expiry/rejection, audit trail/completion certificate/P12 sealing and API operation. It explicitly excludes legal certification, signature-law compliance guarantees, PDF body editing, OCR, redaction, accessibility remediation and general PDF protection. version is 2.41.3 (21 August 2026). Owner remains OpenSign Labs; licence is AGPL-3 with customRoute/third-party carve-outs rather than a simple SPDX assertion. The hosted web edition is vendor_cloud/GUI, hosted API is vendor_cloud/http and Docker is customer_server/http; all require an account. Free hosted/self-hosted signing is documented as unlimited, paid advanced/API features are referenced but no readable price was found; priceFromGbp stays null, hasFreeTier true, pricingModel freemium. dataResidency is self_hosted for the self-hosted edition only; the cloud location is unknown. processesLocally false, requiresAccount true, openSource true, selfHostable true. The two README claims remain current and verified. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  357. d4be689

    The OpenSign dataResidency field and processing note

    What it said
    dataResidency was self_hosted even though the record says the vendor cloud storage location is unverified, so the product-level residency field contradicted the edition-specific note.
    What it says now
    dataResidency is unknown. The self-hosted Docker edition keeps documents on customer infrastructure and that fact is retained in the edition/processing note, but the hosted cloud location is not established, so it is not generalized to the whole product.

  358. f639d3f

    The tesseract.js record, in the OCR job, version, summary, watchOutFor, editions, interfaces, processing/locality and limits

    What it said
    The OCR job was an unresolved prose-only candidate. version was absent despite npm/GitHub release 7.0.0, the web/self-hosted editions had unknown processing and no sources, and the record did not fully define the OCR boundary against PDF support, searchable-PDF output, editing, redaction, signing, accessibility and model-improvement claims.
    What it says now
    The OCR job is scoped to Tesseract.js OCR of images and video frames via a WebAssembly worker in browser or Node.js, with optional word/symbol geometry, hOCR or TSV outputs when enabled. It explicitly excludes PDF reading/rasterisation, searchable-PDF creation, PDF editing, redaction, digital signatures, accessibility remediation/validation, handwriting guarantees, layout reconstruction and hosted-service/model-improvement claims. version is 7.0.0 (latest GitHub release 15 December 2025), Apache-2.0, from the naptha project. Both editions are in_process, browser/library, no account, no hosted service. platforms web/self_hosted; dataResidency self_hosted; processesLocally true with the language-data CDN caveat; requiresAccount false; openSource true; selfHostable true; priceFromGbp null; hasFreeTier true. The two existing claims remain current and verified. qualityScore remains null; launchedYear stays 2015; lastReviewedAt/page-lastmod remain unknown.

  359. 4e5cd3b

    The LEADTOOLS PDF SDK record, in the OCR job, version, licence/pricing/free-tier fields, summary, watchOutFor, platforms, interfaces, editions, processing/locality and claims

    What it said
    The OCR job was an unresolved prose-only candidate even though the PDF SDK page only says OCR can be combined and the actual OCR page is a separate Apryse OCR add-on. version was absent despite help version 23, the seven platform editions had unknown processing and no sources, and the deployment/acquisition sentences were marked verified despite marketing superlatives and no independently built targets.
    What it says now
    The OCR job is scoped to the separate Apryse OCR add-on to the Apryse Server SDK, not the LEADTOOLS PDF SDK: image/image-based PDF input, searchable PDF and structured text output with position data, 80+ languages, customer-side processing, separate Handwriting ICR and Smart Data Extraction, and no included-in-PDF-SDK/free-tier/hosted-service claim. The PDF SDK itself remains a library for PDF creation/manipulation, conversion, extraction, annotations, encryption, signatures, viewing, merging and compression. version is 23; owner remains Apryse with LEAD Technologies noted as acquired. platforms are Windows/Linux/macOS/Android/iOS/web/self_hosted; interfaces library; all editions in_process with no runtime account, while purchase/evaluation is quote/contact-based. priceFromGbp is null, hasFreeTier false, dataResidency unknown, processesLocally true with a hosted-component caveat. The deployment and acquisition sentences are downgraded to reported marketing; the conversion claim remains verified. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  360. 6aa2af4

    The Notability record, in the view job, version, pricing/free tier, platforms/editions/processing, summary, watchOutFor, source and claim

    What it said
    The primary view job was unresolved with no scope. version was absent despite App Store 16.13, priceFromGbp carried an inferred GBP 5.99/month figure although current official prices are USD yearly, the five platform editions had unknown processing and no sources, and the single marketing claim was still verified.
    What it says now
    view is scoped to PDF import/display, automatic PDF outline/table-of-contents detection and navigation, and note-based annotation viewing, explicitly excluding PDF body editing/reflow, OCR of arbitrary scans, text extraction, conversion/merge, redaction, signing and accessibility validation. version is 16.13 (12 October 2026); owner remains Ginger Labs, Inc. priceFromGbp is null with priceBasis unknown because current official plans are USD yearly (Starter free 5 notes; Lite/Plus/Pro), and the old inferred GBP monthly figure is removed; hasFreeTier true. platforms are ios/android/mac/windows/web with GUI interfaces; iOS/mac processing is in_process, web vendor_cloud, Windows/Android unknown because local-vs-cloud is not established; dataResidency us with local/iCloud vs cloud/AI caveats; requiresAccount remains unknown due to mixed native/cloud use. The slogan claim is downgraded to reported. qualityScore remains null; launchedYear stays 2010; lastReviewedAt/page-lastmod remain unknown.

  361. 028c199

    The Okular record, in the view job, version, owner, summary, watchOutFor, platforms, editions, interfaces and processing

    What it said
    The primary view job was unresolved with no scope. version was absent despite release 26.08.2, owner was empty despite the KDE project, the three desktop editions had unknown processing and no sources, and the record did not define the view boundary against editing, OCR, extraction, conversion, merge, redaction, signatures and accessibility.
    What it says now
    view is scoped to PDF reading/navigation: opening/rendering pages, contents/outline and thumbnail sidebars, bookmarks, text selection/copy, magnifier, page navigation, and inspection of annotations, form fields and embedded signatures (including validity/modification checks). It explicitly excludes body-text/image/layout editing, OCR, text extraction, conversion, merging, redaction, accessibility validation and legal certification; annotation/signature creation remains under edit/sign. version is 26.08.2 (8 October 2026); owner is KDE as the developing community/project, with GPL-2.0-or-later for the app and GFDL-1.3 for docs. platforms are linux/windows/mac with Linux primary, Windows via Microsoft Store and macOS as untested nightlies; interfaces empty (GUI desktop app); all editions in_process, no account and no hosted service. dataResidency self_hosted; processesLocally true; requiresAccount false; openSource true; selfHostable true; priceFromGbp null; hasFreeTier true. The two existing claims remain current and verified. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  362. fd31f1e

    The OneDrive document scanning record, in the merge and edit jobs, owner, licence, platforms, editions, processing/locality, summary, watchOutFor and source

    What it said
    The merge and edit jobs were unresolved prose-only candidates. The record did not state that the merge is same-flow capture combination into one PDF rather than merging arbitrary existing PDFs, or that the edit is scan-image crop/rotate/filter plus iOS add-text/highlight rather than editing existing PDF content. Owner and licence were empty, editions had unknown processing, and the product boundary against Lens/SharePoint was not fully documented.
    What it says now
    merge is scoped to combining multiple captures into a single PDF during the OneDrive scanning flow (Android max 10 pages/images in 6.5+, iOS max 30 in 11.31.5+), not merging arbitrary existing PDFs or inserting pages into an existing PDF. edit is scoped to crop/rotate/filter and iOS-only add-text/highlight on the captured scan image before save, not editing an existing PDF body text/images/layout, redaction, OCR, extraction, signing or accessibility. Owner is Microsoft Corporation; licence is proprietary free app/service with OneDrive storage quotas. platforms remain ios/android; interfaces empty (GUI); both editions are vendor_cloud because scans auto-save to OneDrive and no local-save option is documented. processesLocally false; dataResidency multi; requiresAccount true; hasFreeTier true; priceFromGbp null. The two Lens retirement claims remain current and verified. The OCR scope remains rolling-out/export-untested and is not broadened. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  363. 4dca669

    The OpenDocScan record, in the OCR job, summary, watchOutFor, editions, interfaces, processing/locality, source and limits

    What it said
    The primary OCR job was unresolved with no scope. The record had no edition row, interfaces were empty without an explicit GUI/processing explanation, and the OCR boundary was not stated as English-only on-device recognition with a Latin-only text layer rather than a general OCR/extraction or PDF-editing capability.
    What it says now
    ocr is scoped to on-device English OCR with a bundled model and invisible Latin-only Helvetica/WinAnsi searchable-PDF text layer, including confidence/dictionary gating that leaves failing pages without a text layer. It explicitly excludes CJK/handwriting OCR, generic extraction APIs, existing-PDF editing, redaction, signing, accessibility validation and hosted processing. The record has one browser web edition, in_process, empty GUI interfaces, no required account, no hosted service. priceFromGbp null; hasFreeTier true; openSource true; selfHostable true; dataResidency unknown (local browser); processesLocally true with the page-view counter/optional-account distinction. The browser app is the complete product; native Android builds are early. version remains null because no tagged release is published; owner remains null because no company/legal entity is documented; launchedYear remains null. The two existing claims remain current. qualityScore remains null; lastReviewedAt/page-lastmod remain unknown.

  364. 90b28df

    The pdfaPilot record, in the accessibility and convert jobs, version, editions, platforms, interfaces, processing/locality, pricing/free tier and claim

    What it said
    The accessibility and convert jobs were unresolved prose-only candidates. The record had a remote vendor_cloud/http edition for customer-managed Server/CLI workflows, version was absent despite 17.0.683, the Desktop edition included Linux and the Server edition omitted Linux, and the claim was verified without a fresh evidence file.
    What it says now
    Accessibility is scoped to machine-verifiable PDF/UA validation/correction plus PDF-to-HTML structure/tagging views for human review, explicitly not complete semantic accessibility or legal certification. convert is scoped to PDF/PDF-A conversion, many office formats and emails with attachments to PDF/A, and tagged-PDF-to-HTML review, not arbitrary conversion/OCR/extraction/editing/redaction/signing or hosted processing. version is 17.0.683 (12 May 2026); owner callas software GmbH. Desktop is Windows/macOS in_process; Server/CLI/SDK is customer_server on Windows/Linux/macOS with CLI/library interfaces. platforms windows/mac/linux/self_hosted; interfaces cli/library; priceFromGbp null; hasFreeTier false with a free 14-day trial; dataResidency self_hosted; processesLocally true with customer-managed caveat; requiresAccount false. The single claim is downgraded to reported because no fresh evidence file was archived. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  365. 60abd94

    The Reducto record, in the extract and OCR jobs, pricing/free tier, editions, interfaces, processing/locality, summary, watchOutFor and claims

    What it said
    The extract and OCR jobs were unresolved prose-only candidates. The record did not define the API outputs/boundaries, characterised the r-1 1c claim and SOC2/HIPAA as verified, used freemium/free-tier semantics for a pay-as-you-go credit, and had editions with unknown processing, interfaces and sources.
    What it says now
    extract is scoped to parse/extract/split/classify structured output with layout and bounding-box citations, not general PDF editing, searchable PDF, redaction, signing, accessibility or guaranteed field correctness without human review. OCR is scoped as a component of parse (multilingual OCR, page rotation, layout/bbox), not a standalone OCR API, searchable-PDF converter or handwriting guarantee. r-1 is documented as preview; version remains null because no product release version is exposed. owner Reducto, Inc.; licence commercial. Standard is pay-as-you-go with a usage credit, not a permanent free tier: hasFreeTier false, pricingModel paid, priceFromGbp null. platforms api/self_hosted; interfaces http; hosted API vendor_cloud, Enterprise VPC/on-prem customer_server; processesLocally false by default; dataResidency unknown because EU/AU endpoints and on-prem are plan-specific. Both claims are downgraded to reported: the 1c all-in claim is endpoint-specific marketing for a preview model, and SOC2/HIPAA is a vendor/plan statement. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  366. 7f47a6f

    The RoboBraille record, in the convert job, summary, watchOutFor, terms, editions, interfaces, processing/locality and source

    What it said
    The convert job was unresolved with no scope. The record did not precisely enumerate the documented output formats/languages, the free-vs-institutional terms and retention, or the boundary between accessible-format conversion and PDF editing/certification.
    What it says now
    convert is scoped to documented alternate-format conversion/OCR: Braille (contracted/uncontracted, 18 languages), MP3/DAISY, EPUB/Mobi/Kindle/EPUB3, and accessible formats including tagged PDF, DOC/DOCX, Word XML, XLS/XLSX, CSV, TXT, RTF and HTML from images, PDFs and Office documents. It explicitly excludes general PDF editing, extraction APIs, redaction, signing, accessibility certification, guaranteed OCR/Braille/audio accuracy and arbitrary conversion. Free for individual non-commercial users not affiliated with an institution obligated to provide support; no registration; institutional use requires a Sensus service agreement/subscription and no price is published. Hosted in Denmark; source documents deleted after processing, target files retained up to seven days; no API documented (email/web forms only). owner Sensus ApS; platforms web; interfaces empty; one web vendor_cloud edition; dataResidency eu; processesLocally false; priceFromGbp null; hasFreeTier true; selfHostable null; openSource false. The two existing claims remain current. version remains null because no product release version is published; launchedYear stays 2004. qualityScore remains null; lastReviewedAt/page-lastmod remain unknown.

  367. 506d36f

    The SensusAccess record, in the convert job, licence/free terms, summary, bestFor, watchOutFor, edition, processing/locality and source

    What it said
    The convert job was unresolved with no scope. bestFor had a grammar break, hasFreeTier was null despite the institution-agreement/demo path, and the record did not enumerate the documented output formats, OCR/remediation and translation boundaries versus unrelated Remediator/SensusComply services.
    What it says now
    convert is scoped to MP3/DAISY audio, EPUB/EPUB3/Mobi e-books, digital Braille, digital large print, BeeLine Reader, OCR/remediation of image-only PDFs/JPG/PowerPoint/LaTeX and machine translation through the hosted self-service; it excludes general PDF editing, extraction APIs, redaction, signing, PDF/UA/WCAG validation, accessibility certification and output-quality guarantees. The three branded services are kept separate: self-service converter, high-volume Remediator and manual SensusComply. Institutional use requires a monitored service agreement; ordinary users need not register; no public price is published; hasFreeTier is false (demo/trial only) and requiresAccount remains null because API/LMS credentials differ from ordinary self-service use. owner Sensus ApS; platforms web; interfaces http; one web vendor_cloud edition; dataResidency eu (Denmark); processesLocally false; priceFromGbp null; selfHostable null; openSource false. The two existing claims remain current. version remains null because no product release version is published; launchedYear stays 2010. qualityScore remains null; lastReviewedAt/page-lastmod remain unknown.

  368. c265d89

    The SignWell record, in the sign job, pricing/free tier, editions, interfaces, processing/locality, summary, watchOutFor and claims

    What it said
    The sign job was unresolved with no scope. priceFromGbp carried a synthetic GBP conversion of a USD monthly plan, the web/api editions had unknown processing and no sources, and the 65,000-business claim was stale (current page says 80,000+) and still marked verified.
    What it says now
    sign is scoped to uploading/templating PDFs, placing signature/text/date/initials fields, adding recipients and signing order, reminders/conditional rules, browser signing without recipient accounts, completion/audit tracking and API/CLI/MCP/white-label embed routes; it excludes general PDF editing, OCR, extraction, conversion, redaction, accessibility validation and legal certification. owner Docsketch, LLC; platforms web/api; interfaces http/cli; web vendor_cloud (GUI), API vendor_cloud (http/cli); hosted on AWS/US; processesLocally false; dataResidency us; requiresAccount true for senders; priceFromGbp null because no GBP price is verified (the prior figure was a synthetic conversion); priceBasis unknown; hasFreeTier true (1 sender, 1 template, 3 documents/month, API included). Current paid rates are Light annual/ monthly and Business /; Enterprise custom. The 65,000 claim is replaced by current 80,000+ wording and downgraded to reported, as is the no-recipient-account claim because no fresh evidence file was archived. qualityScore remains null; launchedYear 2019 remains founder/about evidence; version remains null because no product release version is published; lastReviewedAt/page-lastmod remain unknown.

  369. 65d61c6

    The Solid Framework record, in the convert job, discontinued note, owner, summary, watchOutFor, interfaces, processing/locality and claims

    What it said
    The convert job was an unresolved prose-only candidate and was not explicitly marked historical even though the product is discontinued. The record risked implying a current standalone conversion product, and the job needed its primary/quality structure preserved after the status correction.
    What it says now
    convert is scoped as historical/legacy PDF-to-Word/Excel/PowerPoint reconstruction and PDF/A archiving via desktop/batch/.NET, explicitly not a current standalone or purchasable capability; new integrations should evaluate the Apryse SDK. The record keeps discontinued true, owner Apryse, one primary job with the existing signpost quality judgement of 68, platforms self_hosted, interfaces library, a historical processing note, unknown data residency, null price/free-tier and the two current claims. It states that no current version, price, OS matrix, formal discontinuation date or wind-down terms are published, and does not claim current OCR/redaction/signing/accessibility/hosted processing.

  370. 3fcca27

    The Surya record, in the extract and OCR jobs, version, summary, watchOutFor, platforms, interfaces, editions, processing/locality and claims

    What it said
    The extract and OCR jobs were unresolved prose-only candidates. version was absent despite PyPI/GitHub release 0.22.1, the four editions had unknown processing and no sources, and the record did not separately define layout/table extraction from OCR or state that Surya does not create searchable PDFs.
    What it says now
    extract is scoped to layout-aware JSON/HTML blocks with text, layout labels, reading order, bounding boxes/polygons, table rows/columns and math; OCR is scoped separately to text-recognition JSON/bboxes. Both exclude searchable-PDF creation, PDF editing, redaction, signing, Office conversion, accessibility validation, hosted API use and guaranteed accuracy. version is 0.22.1 (20 July 2026), Python >=3.10; owner Endless Labs dba Datalab; platforms linux/windows/mac/self_hosted; interfaces library/cli; all editions in_process with local vLLM/llama.cpp inference by default and SURYA_INFERENCE_URL (remote server) caveat; dataResidency self_hosted; processesLocally true with configuration caveat; requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The code is Apache-2.0 and weights are OpenRAIL-M with a commercial threshold/competing-products restriction. The two existing claims remain current. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  371. 25c3ea3

    The tc-lib-pdf record, in the convert job, version, owner/licence, summary, bestFor, watchOutFor, platforms, interfaces, edition, processing/locality and claims

    What it said
    The convert job was unresolved with no scope, and the platforms-vs-prose candidate came from browser wording describing a non-dependency. version was absent despite release 8.81.1, the self_hosted edition had unknown processing and no source, and the record did not enumerate the documented generation/composition outputs versus OCR/extraction/editing boundaries.
    What it says now
    convert is scoped to PDF generation/composition: HTML/CSS/SVG rendering, PDF/A (1/2/3 a/b/u), PDF/X, PDF/UA (1/2), annotations/forms/encryption/PAdES signing, barcodes/images, and PDF page import as Form XObjects. It excludes OCR, text extraction, redaction, arbitrary editing of existing PDF bodies, Office/other-format conversion, independent validation/certification, hosted conversion and a drop-in TCPDF API claim. platforms stay self_hosted; the prose now says rendering engine/service instead of browser. version is 8.81.1 (17 October 2026), PHP ^8.2, LGPL-3.0-or-later; owner Tecnick.com; interfaces library; self_hosted edition in_process; dataResidency self_hosted; processesLocally true with configured-external-interaction caveat; requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The two existing claims remain current. qualityScore remains null; launchedYear stays 2004 from copyright-range evidence; lastReviewedAt/page-lastmod remain unknown.

  372. 720d00e

    The Tesseract OCR record, in the OCR job, version, owner, summary, watchOutFor, platforms, interfaces, editions and processing/locality

    What it said
    The OCR job was unresolved with no scope. version was absent despite release 5.5.3, owner was empty despite the tesseract-ocr project, a hypothetical hosted-integration edition was invented from a possible developer deployment, and the record did not precisely separate image OCR/PDF-output from PDF input/parsing or wrapper/hosted services.
    What it says now
    ocr is scoped to image OCR with Tesseract/lib and the CLI: PNG/JPEG/TIFF/JP2/GIF/WebP/BMP/PNM inputs, tessdata languages (100+), and plain text/hOCR/TSV/ALTO/PAGE/PDF/invisible-text PDF outputs including searchable PDF from an image. It explicitly excludes PDF input/parsing, PDF body editing, redaction, signing, generic format conversion, handwriting/barcodes, hosted OCR services, wrapper behavior and accuracy/compliance guarantees. The hypothetical hosted edition is removed; Tesseract itself has one locally installed engine edition, in_process, library/cli, no account. version is 5.5.3 (24 July 2026), Apache-2.0; owner tesseract-ocr project; platforms windows/mac/linux; dataResidency self_hosted; processesLocally true; requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The one existing claim remains current. qualityScore remains null; launchedYear stays 2005; lastReviewedAt/page-lastmod remain unknown.

  373. ffc06f6

    The Unstructured record, in the extract job, version, owner, summary, watchOutFor, platforms, interfaces, editions, processing/locality and claims

    What it said
    The extract job was unresolved with no scope. The record mixed the open-source library with a hosted-API interface/edition shape, had no version despite 0.27.28, owner was empty despite the privacy policy naming Unstructured Technologies, Inc., and the telemetry/locality boundary was not fully precise.
    What it says now
    extract is scoped to the local OSS partition/extraction library: PDFs, images, HTML, Word/Office, email, EPUB and other supported documents into typed elements, with auto/fast/hi_res/ocr_only PDF/image strategies, optional table inference, coordinates/page breaks and OCR backends. It excludes searchable-PDF creation, PDF editing, redaction, signing, accessibility validation, hosted Transform/Pipelines/API use, inherited hosted terms and accuracy guarantees. The record is the OSS library only: one self_hosted in_process library edition, library interface, no hosted API edition; hosted products are named separately in prose. version 0.27.28 (16 October 2026), Python >=3.11,<3.14, Apache-2.0; owner Unstructured Technologies, Inc.; platforms self_hosted/linux/windows/mac; dataResidency unknown; processesLocally true with metadata-only telemetry caveat (disable via DO_NOT_TRACK/SCARF_NO_ANALYTICS); requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The two existing claims remain current. qualityScore remains null; launchedYear remains null; lastReviewedAt/page-lastmod remain unknown.

  374. 22fac43

    The veraPDF record, in the view job removal, version, owner/licence, Java/runtime, platforms, interfaces, editions, processing/locality and the Java-version prose

    What it said
    The view job was unresolved and semantically overstated what veraPDF does: it is a validator/inspector whose GUI shows validation results/reports, not a general PDF viewer or renderer. version was absent despite release 1.30.3, the four OS editions had unknown processing and no sources, and the Java version range was written as slash-separated numbers that the I18N checker read as ambiguous dates.
    What it says now
    The view job is removed. The record retains accessibility (primary: PDF/A and PDF/UA validation/reporting, policy profiles, feature extraction) and edit (narrow metadata repair only), which are the documented behaviors. veraPDF is explicitly not a general viewer/renderer, converter, content editor, redaction/signing tool or legal-accessibility certification. version is 1.30.3 (9 October 2026), Java 8, 11, 17 or 21, dual GPL-3.0-or-later/MPL-2.0; owner veraPDF Consortium; platforms windows/mac/linux/self_hosted; interfaces library/cli; editions in_process, no account; dataResidency self_hosted; processesLocally true; requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The Java range is comma-separated to avoid slash-date ambiguity. The two licence claims remain current. qualityScore remains null; launchedYear stays 2015; lastReviewedAt/page-lastmod remain unknown.

  375. 8a62742

    The zathura record, in the view job, version, summary, watchOutFor, platforms, edition, processing/locality and source

    What it said
    The view job was unresolved with no scope. version was absent despite 2026.10.4, the Linux edition had an incomplete source/read date, the platform prose mentioned Linux/BSD while the machine field had only Linux without explaining the schema gap, and the view/navigation vs plugin/backend boundary was not explicit.
    What it says now
    view is scoped to local viewing/navigation with zathura and its PDF plugin: page navigation/scroll/zoom, search, index/outline, bookmarks/quickmarks, link following, SyncTeX, automatic reload, rotation/recolour, dual-page/presentation/fullscreen, metadata, encrypted documents, printing and image/attachment export. It explicitly excludes editing, annotation persistence, OCR, extraction service, redaction, signing, conversion and accessibility validation. version is 2026.10.4; owner remains null as a community pwmt.org project; licence zlib core/poppler plugin (other plugins unread); platforms stays linux because BSD has no schema slug and Windows/macOS are not project-supported, with the prose recording Linux/BSD targeting and the downstream macOS formula. One Linux in_process edition; dataResidency self_hosted; processesLocally true with external-link-browser caveat; requiresAccount false; priceFromGbp null; hasFreeTier true; openSource true; selfHostable true. The three existing claims remain current. qualityScore remains null; launchedYear stays 2009 from copyright evidence; lastReviewedAt/page-lastmod remain unknown.

  376. fa3a319

    The Zathura public prose and platform wording

    What it said
    After the view scope and Linux platform were recorded, public prose still named BSD, Windows and macOS (and the external browser) while the schema platform list is Linux-only, so the platforms-vs-prose candidate remained unresolved even though the machine field was correct.
    What it says now
    Public prose names only Linux as the supported machine platform and says other desktop platforms are downstream packaging rather than project-supported builds. The BSD targeting and no-Windows/macOS support remain in non-prose notes. The external-link sentence says external application rather than browser. platforms stays linux; no unsupported platform was added.

  377. 7a053ac

    All eleven task pages

    What it said
    Each page reported a redaction count and a processing complement: what the rest of the index does, rather than what the job needs. On the merge page that answered a question nobody merging documents had asked, and the complement sentences claimed behaviour on behalf of tools whose records said something else.
    What it says now
    Each page states what the job takes and produces and what limits it, and the limit line names no figure because limits are set per tool and the records carry the published ones. The complement sentences are gone, and the remainder is broken into named parts where it is stated at all.

  378. 7a053ac

    All ten comparison pages

    What it said
    Two problems across the set. The short answer and the deciding difference restated each other, so the judgement was a paraphrase rather than a conclusion; and several pairs asserted things their own records did not support, including test coverage being treated as though it established something about privacy, and paid tiers being described as universally uncapped.
    What it says now
    Each answer gives the fact the reader searched for, each deciding paragraph gives the judgement that follows, coverage is scoped to the surface actually tested and dated, and the claims that outran their records are removed. A check now asserts that no deciding paragraph is a paraphrase of its answer above it.

  379. 7a053ac

    The Smallpdf record, in the tagline, the worth-knowing note and the retention sentence

    What it said
    Three blanket statements. The tagline called the product Swiss-hosted, which described where the company is as though it were where files go. The worth-knowing note said every document is uploaded, which described a platform rather than the upload a tool performs. And the one-hour deletion sentence was published without saying whose claim it is or what it covers.
    What it says now
    The tagline says built in Switzerland, the vendor’s wording about the company. The note says using a web tool uploads the file and that whether that suits a confidential document is the owning organisation’s call. And the retention position is the notice’s own three cases, read in a browser rather than inferred: deleted within an hour when signed in unless saved, a saved file deleted within fourteen days of a request, and no fixed period at all for anyone not signed in, including a signature request, where the clock restarts every time the file is reopened.

  380. 7a053ac

    The downloadable dataset, in tools.json and the spreadsheet

    What it said
    The export worked out where files are processed from one input while the pages worked it out from the edition rows as well, so 24 records published one answer in the downloadable data and a different one on their own page. Every surface was internally consistent, which is exactly why nothing caught it: not the build, not the type check, and no existing check.
    What it says now
    The export follows the same order and the same conditions as the site, adds the mixed and self-hosted states, and carries the scope and evidence the records gained. A check now compares the export’s verdict against the badge on the built page for every record, and its value was proved by putting the old flattening back, which failed the check, and removing it, which passed.

  381. 7a053ac

    The SnackPDF comparison page, near the top

    What it said
    The comparison recommended a product built by the person who runs this site, and the disclosure of that sat on the product record and nowhere on the page doing the recommending. A reader deciding between two tools had no way to know from that page that one of them was ours.
    What it says now
    A disclosure field on the comparison itself, rendered in a labelled block near the opening, so a disclosure that belongs on a recommending page cannot be left behind on a record. It also separates unequal test coverage from privacy, because one is a fact about our work and the other is a fact about the tools.

  382. 7a053ac

    The PDFium record, in the summary and the capability tags

    What it said
    It said PDFium cannot create, edit, encrypt or redact PDFs. Its own public API contradicts that: fpdf_edit.h declares FPDF_CreateNewDocument, FPDFPage_New and page-object insertion, so the library can build a document and put objects on a page.
    What it says now
    The record names what the API does and keeps explicit what does not follow from it, which is a complete redaction tool, an encryption workflow, or anything resembling editing in Acrobat. The generic convert tag is gone, an engine with no converter having nothing to convert with, and an edit tag is added with a scope reading document and page objects through the API rather than an editing interface.

  383. 7a053ac

    The pypdf record, in the limitations note and two comparative claims

    What it said
    It said text extraction was limited by an absence of layout analysis. pypdf documents extract_text with extraction_mode set to layout, and documents visitor functions, so the limitation was not there. Two further sentences called text extraction its weakest area and named two other libraries as the better choice, which were comparative judgements with no benchmark behind them.
    What it says now
    The record says what the library documents, keeps the limitations that are real around complex layout and image-only scans, and describes what each library is built for instead of ranking them.

  384. 7a053ac

    The Tabula record, in the capability tags and the activity note

    What it said
    Three things. It was tagged for OCR although Tabula provides none and an image-only scan needs a separate step first. It was tagged as having a browser interface, when the local web interface the desktop application runs is not a browser JavaScript SDK. And it gave March 2025 as the last commit to either repository, from a release date, which is not evidence of abandonment.
    What it says now
    The OCR and browser tags are gone, the application and tabula-java are distinguished as separately maintained things, and the dated abandonment sentence is removed because the pass behind it had run no repository history check.

  385. 7a053ac

    The pdftk-java record, in the programmatic access row

    What it said
    Programmatic access read None, which contradicted both the description above it and the project’s own readme. pdftk-java is used in scripts, and the readme recommends other libraries for direct Java integration rather than presenting itself as one.
    What it says now
    The row reads CLI, and the note that it is not a general Java library API stays in the prose so the distinction is not lost by the label alone.

  386. 7a053ac

    The OneDrive document scanning record, in the capabilities note

    What it said
    It said the help pages document no OCR, so scans are not made searchable. Microsoft publishes a separate guide for recognising, searching and copying text in scanned PDFs in OneDrive on Android and iOS, marked as rolling out, so the absence was wrong.
    What it says now
    The record describes that workflow, says what it does not establish, which is whether recognition is embedded in the file and whether an export carries a text layer, records that no export was tested, and gains an OCR job entry carrying the workflow qualifier.

  387. 7a053ac

    The fpdf2 record, in the limitations note

    What it said
    It said the library is a drawing library and not an HTML renderer. That is a fair description of the core and an unfair description of the product: fpdf2 documents basic HTML rendering through write_html, with no full HTML5 or CSS support.
    What it says now
    The note reads as a primarily programmatic PDF generator with basic HTML rendering through write_html and no full HTML5 or CSS, and the encryption and signing features are recorded as applying to the document fpdf2 produces rather than to an existing PDF it does not edit.

  388. 7a053ac

    The ReportLab record, in the summary, the licence and the merge capability

    What it said
    The summary credited the open-source toolkit with importing pages from an existing PDF so they can be stamped or assembled. The official documentation places page reuse in PageCatcher, which is part of the commercial PLUS package, so a free-software capability was attributed to the wrong licence.
    What it says now
    The two are separate editions with separate licence components, the summary and the worth-knowing note say which side each capability is on, and the merge job carries a scope naming the boundary rather than leaving a reader to find it.

  389. 7a053ac

    The docling record, in the licence field and the browser tag

    What it said
    The licence field said MIT, which reads as covering the whole project. The readme directs users to individual model licences in the packages they come from, so the label covered the code and silently excluded the optional OCR and vision language weights.
    What it says now
    MIT is scoped to the codebase and the model weights are recorded as a separate licence component. The browser interface tag is also gone, since docling-serve is an HTTP API with no documented client-side SDK, so it is its own edition with an HTTP interface.

  390. 7a053ac

    The Surya and Marker records, in the model licence summary

    What it said
    The summary omitted a material clause. The current model licence also restricts entities offering competing products or services, alongside the revenue and funding conditions and the personal and research exceptions, so a reader could believe the threshold was the only restriction.
    What it says now
    Surya names the competing-product restriction, and Marker keeps the written exceptions rather than treating every use above the threshold as forbidden.

  391. 7a053ac

    The LibreOffice Draw record, in the redaction note

    What it said
    The content removal statement was unattached, so it read as though ordinary rectangles or any PDF export removed underlying information. That is the misconception the whole redaction guide exists to correct, and the record was feeding it.
    What it says now
    The statement is attached to File, Export As, Black and White Redacted Export, the workflow the help distinguishes from marks made while reviewing, and the record says plainly that an ordinary rectangle removes nothing. The status stays untested until exported files have been checked for recovery, metadata and attachments.

  392. 7a053ac

    The SensusAccess record, in the programmatic access row

    What it said
    It read None, while the vendor documents a subscription web API and library and learning management integrations. A reader looking for programmatic access was told there was none.
    What it says now
    The row reads HTTP API, and the note separates API access, which needs an account and credentials, from the self-service form, where the fair processing notice says ordinary users need not sign up.

  393. 7a053ac

    The PAC record, in the accessibility verdict

    What it said
    The standard version limitation existed only in the notes, so the published verdict read as though the tool covered whatever PDF a reader brought. Full support is PDF 1.7, PDF 2.0 is best effort, and full PDF 2.0 and PDF/UA-2 are planned without a date.
    What it says now
    The version boundary appears where a reader meets the verdict, the automated checks, the AI hints and the required human review stay distinct from each other, and the offline assurance is marked as reported rather than network tested.

  394. 7a053ac

    The Amazon Textract record, in the free tier row

    What it said
    The free tier read No categorically. AWS documents a three month introductory allowance for new customers that varies by API and excludes Custom Queries, so a reader planning an evaluation was told there was nothing to evaluate with.
    What it says now
    The record describes that allowance. The free tier field stays false, because the point is that it is a trial rather than an ongoing tier, and related storage and notification charges are named without estimating a bill.

  395. 7a053ac

    The Stirling PDF record, in the processing and test statements

    What it said
    A claimed self-hosted test stood beside an inherited statement that the self-hosted instance had not been checked. Two sentences on one record said opposite things, and a reader had no way to tell which was true.
    What it says now
    The processing statement is deployment specific, files go to the instance you run wherever it is hosted, and the test is scoped to one operation, one version and configuration, and the hosts it captured.

  396. 7a053ac

    The borb record, in the redaction verdict

    What it said
    The badge said it removes content, confidently, while the record’s own evidence contradicted it: the open-source annotation implementation is a placeholder, and the vendor’s two pages disagree about whether the paid path removes content or hides it.
    What it says now
    The verdict is unverified, because hidden and removed are different claims and the evidence supports neither. The telemetry inconsistency in the privacy policy is recorded rather than resolved, and a claim about being the first network-active local library is gone.

  397. 7a053ac

    The Telerik Document Processing Libraries record, in the processing note

    What it said
    It said no document ever leaves the process, which had been inferred from the absence of an Office or Acrobat dependency. Not depending on something is not evidence of no network traffic, and the sentence sat beside a shared disclaimer saying the opposite.
    What it says now
    The library is described separately from the application hosting it, the documented generative AI, timestamp and signing paths are named as configured external routes, merge, sign and protect membership is added, and the record notes that generated output does not automatically pass accessibility testing.

  398. 7a053ac

    The SetaPDF-Core record, in the price row and two capability claims

    What it said
    Pricing read quote-based with no published figure, while the vendor publishes the server licence at EUR 180 excluding VAT, tied to one server or virtual machine, with a separate cluster and cloud licence and renewals at 20 percent. Separately, the record said Core provides no drawing API, and its feature list includes a page canvas object. And pure PHP was treated as a guarantee of local-only deployment, while HTTP writers and S3 stream wrappers are documented.
    What it says now
    The published figure, the renewal rate and the basket effect are recorded, since Core is included in each component so a workflow needs a basket rather than a sum. The drawing claim is removed, and the deployment claim is a statement about where the library runs rather than about what it can be configured to reach.

  399. 7a053ac

    The pdftools PDF Web Viewer record, in the redaction claim

    What it said
    It called itself the only tool in the index documenting client-side redaction. MuPDF.js documents browser-capable redaction too, so the superlative was wrong on a fact a reader could check in the index itself.
    What it says now
    The claim is gone. Client-side and air-gapped deployment are recorded as options rather than proof that every embedding application keeps files local, since the SDK can open URLs, and edit is described as annotation and form operations rather than general editing of existing text.

  400. 7a053ac

    The PDF Expert record, in the processing badge

    What it said
    The badge said no server is involved, while the record’s own note described conversion uploads, AI providers and optional cloud sync. The badge is the part a reader sees first and the caveat was underneath it, so the two most visible things on the record disagreed.
    What it says now
    The badge reads Mixed, depends on version and feature. Installed editing stays separate from conversion uploads, AI providers and enabled cloud sync, OCR is scoped to Mac 3.0 and later with iOS not yet available, and the lifetime licence being Mac-only is explicit.

  401. 7a053ac

    The SimplePDF record, in the why-no-redaction note

    What it said
    The negative rationale said the vendor had no whiteout mention, which was the ground for a no-redaction verdict. The vendor’s official guide for the whiteout tool exists, so the reason given for the verdict was untrue even though the verdict happened to be right.
    What it says now
    The guide is cited, and the verdict stays no redaction on the ground that actually supports it: a whiteout covers text, and secure removal has not been established.

  402. 7a053ac

    The SumatraPDF record, in the redaction verdict

    What it said
    It said there is no redaction at all. The vendor documents applied redaction in the 3.7 pre-release, so a product-wide removal claim was wrong and it was wrong in the direction that reads as reassuring.
    What it says now
    The verdict is unverified, with the version boundary stated, rather than a claim covering a whole product.

  403. 7a053ac

    The iLovePDF record, in the processing and redaction notes

    What it said
    The notes shouted. Emphatic capitals and the internal history of which correction changed which word were published as though a reader needed them, so the evidence a reader could use was buried under the account of how it had been arrived at.
    What it says now
    Short neutral evidence notes, dated where a date matters. Two capitals deliberately stay: one is a literal token a check reads rather than emphasis, and the other is an acronym.

  404. 7a053ac

    The jsPDF record, in the processing note

    What it said
    It claimed no network call exists in the library, as an absolute. That is close to true and not true: the browser file-loading module performs GET requests for resources the caller names, so a reader could configure a fetch the sentence denied.
    What it says now
    Generation needs no vendor-hosted rendering service, and the caller-named resource loading is stated, with the distinction that loading a resource is not uploading a document.

  405. 7a053ac

    The RoboBraille record, in the range claim and the free-use condition

    What it said
    Its Braille output was described as broader than anything else in the index, a comparative claim with no comparison behind it. And the free-use condition was narrower than the word free suggests: individual, non-commercial users not affiliated with an institution obliged to provide support, with institutions needing a subscription.
    What it says now
    The superlative is gone. The free-use condition appears where a reader meets it rather than only in the notes, and the documented library and learning management integration is described as an institutional arrangement rather than a public API.

  406. 7a053ac

    The Adobe PDF Accessibility Auto-Tag API record, in the positioning claim

    What it said
    It said this was the only tool in the index tagging at scale through an API. Equidox documents API-integrated batch remediation and is already listed here, so the superlative was wrong on a fact a reader could check in the index.
    What it says now
    The claim is gone. The warning stays, that it replaces existing tags and does not guarantee WCAG or PDF/UA compliance.

  407. 7a053ac

    The Allyant PDF Validator, CommonLook Office and AccessibilityNow records, in the requirements and tone

    What it said
    Three things across the records. An Acrobat X version floor was stated as fact while the current page calls the product an Acrobat Pro plug-in. Phone contact details were marked as required when they are not, so the requirements list was stricter than the vendor’s. And a claim that the verification rebuttal was complete was scoped to the automated portion only, while the vendor guides a person through further checkpoints.
    What it says now
    The version floor is marked unconfirmed, the fields are not all mandatory, a plug-in architecture is not treated as evidence of network behaviour in either direction, the rebuttal is scoped to what it covers, and marketing built on fear is replaced by the vendor’s claims attributed with the questions a buyer should ask.

  408. 7a053ac

    The Amazon Textract, Google Document AI, Adobe PDF Extract and Azure Document Intelligence records, in the file size rows

    What it said
    Each carried a bare maximum size number, which is the wrong shape for a cloud API: the limit depends on whether the call is synchronous or asynchronous, what the input is, and how many pages it has. A single figure told a reader a limit that did not apply to their call.
    What it says now
    Mode-specific limits, every figure taken from the vendor’s documentation: Textract at 10 MB synchronous and one page for PDF or TIFF, 500 MB and 3,000 pages asynchronous, and 10 MB either way for JPEG and PNG. Google Document AI at 40 MB online and 1 GB batch, with 30 contiguous pages for eligible imageless online requests. Adobe at 100 MB, 400 pages native and 150 scanned, kept nominal because complex files can hit lower limits. Azure’s 500 MB scoped to tier, API version and model. The single size field stays empty on all four, because one number is what was wrong.

  409. 7a053ac

    The Google Cloud Vision and Document AI record, in the privacy attribution

    What it said
    The record covered two services as one, and cited a data usage notice belonging to Vision as though it established retention for Document AI. Input support, storage, regions, prices and limits also differed between them.
    What it says now
    The two are kept apart, the linked data usage notice is marked as Vision’s and cannot establish Document AI retention, and the structured extraction use case gained extract membership with a scope naming Document AI processors.

  410. 7a053ac

    The DocuSeal record, in the licence note

    What it said
    It described the additional terms as imposing extra conditions on commercial or competing use. The terms file requires retaining original attribution in interactive interfaces, which is an attribution condition rather than a restriction on competing, so the record overstated what the licence forbids.
    What it says now
    The record says what the condition is, and keeps the AGPL obligations, the attribution requirement and the separately licensed Pro features distinct from each other.

  411. 7a053ac

    The DocuSeal and Documenso records, in the self-hosting claims

    What it said
    Self-hosting was taken to mean every communication stays on your own server. It does not: hosting and storage are deployment choices, and DocuSeal documents local disk or AWS, Google and Azure storage, plus SMTP delivery for the messages it sends.
    What it says now
    The records say what self-hosting decides and what follows from configuration instead, which is the distinction a reader running a confidential workflow actually needs.

  412. 7a053ac

    The Aspose PDF and Spire PDF records, in the evaluation limits

    What it said
    Both asserted that a longer operation silently truncates a real document. The vendors document page limits, at four evaluation pages for one and ten for loading with three for image conversion, and a documented page limit is not the same finding as a failure mode. Neither was tested.
    What it says now
    The documented limits are kept, the words trap and worse are gone, and the requirement to use a temporary licence and test the full document is recorded.

  413. 7a053ac

    The Syncfusion PDF record, in the watermark and licence rows

    What it said
    The watermark field conflated the Community licence with an unlicensed trial, so a reader on the free Community licence was told to expect a watermark they would not get. The record also carried a browser tag on an in-process library.
    What it says now
    The Community licence is its own route with its own eligibility stated: under one million dollars gross revenue, at most five developers and ten employees, no more than three million outside capital, plus the vendor’s conditions for client and government work. The browser tag is gone.

  414. 7a053ac

    The PDFlib record, in the processing note and the task scope

    What it said
    It said documents are generated and never sent anywhere, which is an architecture described as a network test. The vendor’s own feature matrix includes internet download helpers and network resources. The task scope also missed documented tagged PDF, PDF/UA-1 and encryption support.
    What it says now
    The statement is scoped to an embedded library rather than to observed network behaviour, accessibility and protect membership is added from the documented features, and the paid levels for PDI and PPS are kept distinct from the base product.

  415. 7a053ac

    The ICEpdf record, in the maintenance narrative

    What it said
    It read as an abandoned project, built on the state of the original vendor site. The community project on GitHub is maintained, documents Apache 2.0 and Java 11, and publishes current artifacts.
    What it says now
    A neutral distinction between the legacy site and the maintained community project, with the licence and current version recorded. A competitor’s marketing is not used as evidence of maintenance or support.

  416. 7a053ac

    The pdftools SDK record, in the hosting claim and the outlook

    What it said
    It said the same company sells a hosted conversion service, which the documentation does not establish: self-hosted, on-premises and cloud deployment on Windows, Linux and Docker are all documented, and a REST API is not evidence that the vendor hosts it. The record also predicted that the acquisition would change commercial terms.
    What it says now
    The documented deployment options, with the note that a REST API proves nothing about hosting, and the documented acquisition notice in place of a prediction. Redaction stays uncertain for this SDK rather than borrowing the viewer product’s verdict.

  417. 7a053ac

    The GroupDocs record, in the price row and the redaction warning

    What it said
    Pricing read quoted with no figure, while the vendor publishes dated starting prices. And the redaction warning stopped at hide or remove, when the .NET documentation describes rasterised PDF as the default save path, which loses searchable text and original metadata.
    What it says now
    The published starting prices are recorded as starting configurations, and the redaction warning carries the output mode and what it costs in searchable text and metadata, with original-format output named as the option that avoids it.

  418. 7a053ac

    The pdfRest record, in the price row and the usage note

    What it said
    It said no figure is published, while the vendor publishes a Cloud, AWS and Container matrix, with Cloud Starter at 100 calls a month and watermarked output. A reader also could not tell what a call is: billable calls include POST, DELETE and status polling rather than one per document.
    What it says now
    The matrix with its starting tier and watermark, and an explanation of what gets billed, which is the difference between a cheap-sounding price and an actual monthly cost.

  419. 7a053ac

    The Goodnotes record, in the processing note, the watermark and the free plan limit

    What it said
    It said files stay on your device, scoped to the web. That flattened Apple-local storage, iCloud and Goodnotes Cloud into one answer, marked free exports as clean when the vendor documents a watermark, and gave a file size limit without saying it applies to the free plan only.
    What it says now
    An edition and feature specific description that separates Apple local and iCloud from Goodnotes Cloud, records free exports as vendor-documented watermarked, and scopes the limit to the free plan. A pricing and support conflict in the vendor’s own pages is exposed rather than resolved.

  420. 7a053ac

    The Foxit PDF Editor and Nutrient records, in the processing and licensing notes

    What it said
    Foxit’s local processing claim covered features that are not local: Smart Redact runs as cloud processing on Azure with vendor-reported deletion and a fifty-document allowance. Nutrient’s local badge covered the whole product when it applies to the client-side Web SDK, and its licensing was described as uniform when it is product specific.
    What it says now
    Foxit records that Local AI runs offline while Smart Redact is cloud processing. Nutrient restricts the local claim to the client-side Web SDK, records product-specific licensing, and keeps the redaction limitation that the application must be applied before export and does not sanitise all metadata.

  421. 7a053ac

    The merge task page, in the tool table

    What it said
    The page carried a redaction column and a redaction total, on a task where whether a tool can remove content has nothing to do with combining files. It invited a reader to weigh a criterion the job does not use.
    What it says now
    The column appears only where content removal is a live question for that job and a tool has a finding, so merge has none and edit still does. The same rule is applied from the other side on the task pages, where the counts describe the job rather than the index.

  422. 7a053ac

    The merge task page, in the numbered recommendations

    What it said
    A bare numeral sat before each recommendation, which read as a stray list number that had drifted between unrelated rows rather than as any kind of ranking.
    What it says now
    Each reads Our pick followed by its number, and the page states how the shortlist was chosen, so a labelled recommendation is traceable to a method.

  423. 7a053ac

    The merge task page, in the PDFsam row

    What it said
    The row carried generic product detail on a page about merging, so a reader learned what the product is rather than whether it merges well.
    What it says now
    Merge-specific facts: free installed merging, page ranges, bookmark handling, field-name collision handling and page mixing, with Enhanced redaction kept as a separate paid edition rather than folded into the free listing.

  424. 7a053ac

    The merge task page, in the SumatraPDF entry

    What it said
    The entry presented merging as available without saying that the vendor documents it for the 3.7 pre-release, so a reader on the current stable build would reasonably expect a feature they may not have.
    What it says now
    The entry carries a scope naming the pre-release, so the version boundary is on the row rather than in a note a reader has to go looking for.

  425. 7a053ac

    The SnackPDF and SimplePDF comparison, in the two opening blocks

    What it said
    The short answer and the deciding difference were substantially the same paragraph, so a reader got one idea twice and the judgement that was supposed to follow the fact was buried in the second copy.
    What it says now
    One concise answer states the fact, and the second block covers a different question: everyday local form filling against collecting form responses from a group.

  426. 7a053ac

    The SnackPDF and SimplePDF comparison, in the opening

    What it said
    The opening implied that every SimplePDF workflow keeps files off the device. Its paid submission workflow does not: the vendor says the completed document is stored on their side or in storage you supply.
    What it says now
    The subject is about editing, and the answer separates free local editing from the paid submission workflow, so the local claim is scoped to the part of the product it is true of.

  427. 7a053ac

    The SnackPDF and SimplePDF comparison, in the price row

    What it said
    It showed only a rounded figure converted into pounds, which made a free personal editor look paid and hid which plan the number belonged to.
    What it says now
    The vendor’s own figure in its own currency appears first with the conversion labelled, and the record holds it separately so a comparison can prefer the published number over a derived one.

  428. 7a053ac

    The Tesseract and OCRmyPDF comparison, in the dependency claim

    What it said
    It said OCRmyPDF inherits Ghostscript as a dependency. The current documentation lists pypdfium2 as an alternative, so Ghostscript is not strictly required, though older and packaged builds differ.
    What it says now
    The dependency is qualified rather than stated as inherited, which matters because the licence you are bound by depends on which of the two you have.

  429. 7a053ac

    The OCRmyPDF record, in the licence summary

    What it said
    The summary implied Ghostscript is always bundled, so a reader assessing licence obligations would count a dependency that may not be there.
    What it says now
    The core licence is matched to the repository and the dependencies are qualified, because which licences bind you depends on what you install or distribute.

  430. 7a053ac

    The PDFsam record, in the free, open-source, account and platform fields

    What it said
    Four fields mixed three editions together, so a reader could not tell which edition a badge described and reasonably assumed the answers applied to the product as a whole.
    What it says now
    Three edition rows carry their own platforms, licences and processing notes, redaction is recorded as an Enhanced capability holding its own plan evidence, and the merge listing names Basic so the free path is not confused with the paid one.

  431. 7a053ac

    The Tesseract OCR record, in the processing statement

    What it said
    It said no server is involved. For an embeddable library that is an absolute statement about something the library does not decide: where it runs depends on how a developer deploys it, and a hosted integration can receive files.
    What it says now
    The deployment distinction, with the engine described as locally installed and its processing location depending on how it is deployed, and the PDF input limitation stated in the same field rather than separately.

  432. 7a053ac

    The MuPDF.js record, in the package, task scope and licence claims

    What it said
    It did not name the official package, so a reader could install a deprecated community one; it carried an unverified bundle size figure as fact; and it made a categorical claim about when a commercial licence is needed.
    What it says now
    The official npm package is named and the deprecated community package is distinguished from it, view, extract and image-output membership is added, the commercial licence is described as the vendor’s own distinction rather than a rule, and the bundle figure is removed rather than carried forward.

  433. 7a053ac

    The Bluebeam Revu record, in the processing badge

    What it said
    The badge was scoped to the web while the record covered a desktop application alongside Studio sessions and cloud apps. Ordinary offline subscription access also needs a Bluebeam ID, with reauthentication every thirty days, which the record did not say.
    What it says now
    The badge reads Mixed, depends on version and feature, and the note explains what the account requirement means for someone who expected a desktop application to work offline indefinitely.

  434. 7a053ac

    The Tesseract and OCRmyPDF comparison, in the availability row

    What it said
    The row was labelled Currently working, and the evidence behind it was a website response time. A vendor’s website answering says nothing about whether an installed application works, and the label asserted more than the measurement could carry.
    What it says now
    The row is named Website responded, so it says what was measured rather than what the measurement was taken to mean.

  435. 7a053ac

    The SimplePDF record, in the vendor figure note

    What it said
    The note speculated that one vendor figure was small enough to be plausible for a tool of this kind, and that no source exists to look for. Both are judgements about a number rather than findings about it, and neither had an audit behind it.
    What it says now
    The figure is reported as the vendor’s, with the method unpublished and the usage unaudited, and whether it looks plausible is not stated at all.

  436. 7a053ac

    The SnackPDF record, in the test findings and the tool counts

    What it said
    Historical test findings were presented in a way that read as tests this pass had performed, and the page carried freshness and count notes that contradicted each other.
    What it says now
    The findings are dated results from the runs performed on 16 September and after, each naming the workflow it covers, with the tool counts reconciled to this date and the vendor material read since supporting the mechanism without claiming to reproduce those tests.

  437. 7a053ac

    The iLovePDF record, in the file size limit

    What it said
    The limit was a universal unknown, which is not what the vendor publishes: the pricing page lists 100 MB for merging and 200 MB for compression on the limited tier, so the honest answer is that the limit is set per task rather than that there is no figure.
    What it says now
    The listing says the limits are set per task and quotes the published ones, and the single maximum size field stays empty rather than carrying one figure across every task.

  438. 7a053ac

    The iLovePDF record, in the summary, the price and the region

    What it said
    The summary said the product works across web, desktop and mobile without installing anything, when the desktop and mobile applications are downloads. The price showed five pounds a month with nothing saying that is the annual rate, with monthly billing at nine. And the data region read European Union while the record’s own note quoted a policy that allows processing outside it.
    What it says now
    The summary says the web tools need no install and the apps are separate downloads under one account, the price carries the commitment behind it, the region reads multiple rather than the narrower half of a two-part statement, and the worth-knowing note separates the tools that need an account from those that do not.

  439. 7a053ac

    The Smallpdf record, in the vendor user count

    What it said
    A vendor figure was met with our own judgement that it is roughly a fifth of the world’s population and therefore not credible. Neither claim had an audit behind it, and a plausibility verdict about a large number is not a finding about the product.
    What it says now
    The attribution is kept and what is missing is stated: the figure is the vendor’s, what it counts is not said, whether people, sessions, documents or visits, and no method or independent audit is published, so it is unverified and we cannot say what it measures.

  440. 7a053ac

    The pdfplumber and pdfminer.six records, in the dependency wording

    What it said
    Both said they bring no compiled extensions and need no compiled system dependencies. pdfplumber requires pypdfium2 and Pillow, and pdfminer.six requires cryptography and optionally Pillow, so the project code being pure Python said nothing about what installing it brings.
    What it says now
    Both records say the project code is pure Python while the installation brings those dependencies, and that wheel availability and source-build prerequisites depend on platform and version.

  441. 7a053ac

    The pdfplumber record, in the scanned page note

    What it said
    It said a scan needs a text or explicit-lines strategy, which is not what a table strategy does with a scan. The sentence pointed at a setting rather than at the missing step.
    What it says now
    Borderless digital tables and scanned pages are stated apart, with recognition named as the step that is absent and no promise made about the quality of what comes out of it.

  442. 7a053ac

    The pikepdf record, in the capability tags

    What it said
    The record carried a generic conversion tag. pikepdf rewrites and linearises a document’s structure, and its own overview distinguishes that from rasterising a page or changing the format, so the tag said more than the library does.
    What it says now
    The conversion tag is gone, leaving the record claiming only the structural operations it documents.

  443. 7a053ac

    The Poppler record, in the command line limitation

    What it said
    It said the command-line tools take one input file per invocation. pdfunite accepts several source documents and one destination, which its own source declares and loops through, so the limitation was simply false.
    What it says now
    The record separates document assembly from batch processing of unrelated jobs, and does not infer a native batch interface from the fact that merging takes multiple inputs.

  444. 7a053ac

    The pdfcpu record, in the content removal statement

    What it said
    It said there is no redaction and no true content removal, which collapsed two different things and overstated the second. The project documents removing stamps, watermarks and signatures, and manipulating pages and objects.
    What it says now
    The absence of a documented general redaction workflow is stated separately from those removal operations, with the note that they are not proof that arbitrary content has been securely removed.

  445. 7a053ac

    The Apache Tika record, in the crash isolation claim

    What it said
    So a malformed file cannot take the host down was published as a guarantee. The vendor describes crash-isolated forked parsing in the 4.x line, which isolates a parser crash rather than conferring immunity to resource exhaustion or to every kind of failure.
    What it says now
    The record describes that isolation for what it is, and dates the version facts it rests on.

  446. 7a053ac

    The MuPDF and mutool record, in the redaction evidence

    What it said
    The redaction evidence cited a different project’s page documentation, which describes an interface this record is not about, so the citation did not support the entry it sat under.
    What it says now
    It cites MuPDF’s own page API, says which interface that is, and notes that the mutool subcommand collection is a different set of operations again. Removal stays a documented capability and independent safety stays untested.

  447. 7a053ac

    The Genius Scan record, in the data region

    What it said
    The region read European Union and the note said EU backend, while the privacy policy permits storage in the EU or in countries the vendor considers adequate under European rules, which is wider. The narrower half of a two-part statement had been published alone.
    What it says now
    The region reads multiple, and the workflows are separated rather than promised together: document sync, exports and analytics have their own destinations. A European company is not treated as a guarantee of European-only processing.

  448. 7a053ac

    The Genius Scan record, in the free tier

    What it said
    The free tier was described as fully functional with the paid features blurred into it, so a reader could believe recognition and cloud backup were included. The watermark field was empty when the vendor’s pricing page states that scans are produced without watermarks.
    What it says now
    The summary separates unlimited basic scanning from the paid features, which are recognition, cloud backup and sync, web access, advanced organisation and extra export targets, and the watermark answer is false on the vendor’s own statement with the citation recorded and no output test claimed.

  449. 7a053ac

    The Adobe Scan record, in the storage note

    What it said
    The phrase rather than staying on the phone implied that cloud backed storage excludes a local export, which is not what the vendor says.
    What it says now
    The record separates cloud-backed storage from export destinations, since the vendor lists local storage alongside the account and connected services, and it says plainly that a local export is not proof that capture or recognition happened offline.

  450. 7a053ac

    The SwiftScan record, in the AI processing claim

    What it said
    It concluded that the AI summarise and translate tools are cloud features, stated as established fact, while the record’s own caveat said only that the vendor does not describe them as running on the device.
    What it says now
    The record says their processing location has not been confirmed, and says why: advertising a feature does not establish where it runs.

  451. 7a053ac

    The Apache FOP record, in the positioning sentence

    What it said
    It called FOP the reference open source implementation of XSL-FO, which is a claim about a standards body that no project page can establish, and it presented XSLT as the way in when XSLT is one common way to produce the input rather than the required one.
    What it says now
    The reference-implementation claim is gone. FOP consumes formatting objects, XSLT is described as a common way to produce that input, and an application may supply the formatting objects directly. The distinction from an HTML and CSS renderer is kept.

  452. 7a053ac

    The Prince record, in the release cadence note and the platform list

    What it said
    It inferred that a fix may wait for the next major version, which the download page contradicts: prerelease builds carrying fixes and improvements are offered, so a stable release date does not describe the support cadence. The record also described its missing operating system list as a product limitation when it is a gap in our reading.
    What it says now
    The inference is removed, the platform list is described as something we could not read rather than something the product does not publish, and the dated release fact is kept.

  453. 7a053ac

    The docling, marker, surya, grobid and tesseract.js records, in the comparative claims

    What it said
    Five records carried phrases like the highest quality Markdown, the most capable open option and the strongest open tool. No comparison had been run, so each was a ranking with nothing behind it, and one rested on a vendor’s own benchmark presented as our conclusion.
    What it says now
    Each says what the tool is built for, and a vendor’s own benchmark is attributed to the vendor rather than converted into a ranking.

  454. 7a053ac

    The Tesseract.js record, in the output formats

    What it said
    It described word and symbol geometry, hOCR and TSV as what the library returns. The readme says version six disables outputs other than text by default and shows hOCR being switched on explicitly, so the description covered what is available rather than what comes back.
    What it says now
    The outputs are qualified by version and described as opt in, and the distinction between image input and no direct PDF support is untouched.

  455. 7a053ac

    The LibreOffice Draw record, in the programmatic access row

    What it said
    Programmatic access read None, which the suite’s own documentation does not support: it documents command-line conversion and export, and the PDF export filter takes documented options.
    What it says now
    The field reads CLI, and the prose says what that covers, which is export automation, and what it does not, which is an embeddable drawing SDK or faithful editing of every imported PDF.

  456. 7a053ac

    The zathura record, in the processing note

    What it said
    Two absolutes the project does not support: no network component described, and nothing leaves the device. Its documentation includes opening links in an external browser, which sends nothing about the document but is an outbound action. The record also called it the most configurable viewer in this index, a comparison defined nowhere.
    What it says now
    The note is scoped to local document viewing and names the external-browser exception, and the superlative is gone.

  457. 7a053ac

    The SensusAccess record, in the comparison and limits

    What it said
    It carried a directory-wide superiority claim with nothing behind it, and treated three branded services as one product, so limits were inherited from one across all of them.
    What it says now
    The three services are named separately, the self-service converter, the high-volume remediator and the manual compliance service, with their limits kept apart. The vendor-reported retention is recorded: source deleted after processing, delivery files up to seven days, and possible storage in spam filtering systems.

  458. 7a053ac

    The Azure AI Document Intelligence record, in the container claim

    What it said
    It said the same cloud models run in containers. The installation documentation supports Read and Layout for one version, with additional models on two earlier ones, so parity is version and model specific. Billing telemetry was also being read as document upload.
    What it says now
    The version and model detail is stated, billing traffic is distinguished from sending a document, disconnected licensing is separated, and the record notes that a container image does not require a Linux host.

  459. 7a053ac

    The Reducto record, in the price row

    What it said
    Pricing read as quoted with no figure, while the vendor publishes per-endpoint prices, including a rate per thousand pages and an introductory allowance.
    What it says now
    The published tiers and the per-endpoint figure are recorded, with the credit explained as the billing unit rather than a page.

  460. 7a053ac

    The DocuSeal record, in the API and embedding costs

    What it said
    The free cloud request allowance was presented without saying which tier it belongs to, and API access and embedding were not distinguished from the plan a reader might already be on.
    What it says now
    The allowance is scoped to the cloud tier, API and embedding are recorded as paid features with a per-completed-document charge including on the self-hosted Pro edition, and the vendor-reported indefinite cloud retention is stated.

  461. 7a053ac

    The tc-lib-pdf and Gotenberg records, in the processing claims

    What it said
    tc-lib-pdf was described as generating in process, which is not the same as having no external interactions: the extensions it needs, optional outbound calls for timestamp authorities, signing and revocation endpoints, and controlled remote resource access all sit outside the process. Gotenberg’s files stay on your device was a claim about a server product.
    What it says now
    Both are deployment specific. tc-lib-pdf names the extensions, the outbound paths and the fact that it is not a drop-in replacement for the project it came from. Gotenberg says files go to the instance the operator runs, and that URL conversion, outbound filtering, downloads and webhooks mean no workflow has zero outbound transfers.

  462. 7a053ac

    The PDFShift record, in the file size and the rendering engine

    What it said
    A single 250 MB maximum was published while the pricing table lists a no file size limit feature on some tier, so the universal figure contradicted the vendor’s own table. And exactly as a browser shows it was doing work it could not do.
    What it says now
    The maximum size field is empty with the free plan figure kept in the note, because one number was the problem, and the rendering is described as Chromium based.

  463. 7a053ac

    The Antenna House Formatter record, in the interfaces

    What it said
    An exclusive command-line label missed the documented routes: the current documentation also covers Java, .NET, COM and C or C++ interfaces.
    What it says now
    Both are recorded. The full and Lite split is also stated, with tagged PDF and PDF/UA marked unavailable in Lite, so the record does not imply that every purchase produces accessible tagged output or that support certifies a document.

  464. 7a053ac

    The Vivliostyle record, in the licence and component claims

    What it said
    It called the project the strongest copyleft in this index and one of the very few, neither of which has a defined comparison behind it, and it treated the project as one thing.
    What it says now
    The superlatives are removed, the viewer, command line, core and hosted editor are separated with their own statuses, the command line is recorded as producing several publication formats rather than PDF only, and the copyleft attribution is tied to each component and version.

  465. 7a053ac

    The Aspose PDF record, in the local processing claim

    What it said
    Local document processing was presented as proof of no external contact. The licensing documentation describes metered key activation and a licence server, which is connectivity about licensing rather than about the document.
    What it says now
    Licensing connectivity is scoped separately from document content transfer, and the vendor’s inconsistent grace periods are deliberately not published as established.

  466. 7a053ac

    The IronPDF record, in the licence counts and the network claim

    What it said
    Licensing was described as chiefly per developer when the vendor counts developers or development machines and defines project and location limits separately, so the count a buyer has to satisfy was wrong. A web-service disclaimer was inherited from a different kind of product, the no-server statement was absolute for a library that can run on a customer’s own server, and an exhaustive claim about its only network calls was published.
    What it says now
    Developer or machine counting, project and location limits, and the redistribution, SDK distribution and air-gapped licensing routes are all recorded. The inherited disclaimer is gone, the no-server statement is qualified, and the network paths are the documented ones.

  467. 7a053ac

    The Qoppa Java PDF SDK record, in the support status

    What it said
    It asserted that patches, security fixes and compatibility updates all stop. That is a sales fact turned into a support fact: a notice that no new licences are sold does not establish what existing licensees receive.
    What it says now
    The support assertion is removed, the related claim that an existing deployment is unaffected is qualified for the same reason, the free single-purpose library is kept separate from the commercial suite, and signing and compression membership is added.

  468. 7a053ac

    The FPDI record, in the limitations note

    What it said
    The limitation was described loosely as compressed cross references and streams, and it was set beside a version number in a way that implied the version established how common the problem is. A version introducing a format does not say how many documents use it.
    What it says now
    The limitation is named precisely, as compressed cross-reference streams and object streams, with the note that the PDF 1.5 introduction does not establish prevalence. What imported pages do not preserve is listed, including the link annotation opt in added in 2.4.0.

  469. 7a053ac

    The PDFsharp record, in the preview claim

    What it said
    It said the library can never produce a preview. That is a statement about the library doing a job it does not attempt, when the accurate statement is that it has no renderer.
    What it says now
    The library lacks a PDF renderer, the preview control for drawing operations is noted separately so the two are not confused, and merge membership is added.

  470. 7a053ac

    The JPedal, Solid Framework and Datalogics records, in the positioning claims

    What it said
    JPedal was described as a browser grade renderer, which language and architecture do not establish. Solid Framework carried tombstone wording implying nobody buys it, which is a sales conclusion the vendor’s pages do not support. Datalogics was called the most expensive by a wide margin, with no comparable quotes behind the comparison.
    What it says now
    The renderer claim is gone, with protect and merge membership added and compression scoped. Solid Framework gives practical guidance with the sales conclusions qualified. The comparison is removed while the published starting price stays, and merge and extract membership is added with compression scoped.

  471. 7a053ac

    The iText, Scanbot and gImageReader records, in the licensing and locality notes

    What it said
    Three overgeneralisations. iText’s tagline called the commercially licensed add-ons one thing, when some are open source and some are closed. Scanbot’s self-hostable answer was a blanket no against documented server editions, and local engine capability was treated as proof that results stay on a device. gImageReader’s images never leave the machine was an absolute about software the user installs.
    What it says now
    The iText add ons are component specific, with the open-source and closed ones named separately. Scanbot records the server editions, revisits the self-hosting answer, and notes that a local engine with configurable routing to server workflows is not the same as results staying put. gImageReader’s statement is the documented local desktop architecture, and its edit membership is defined as correcting recognised text and recognised output.

  472. 7a053ac

    The PDF-XChange SDK and pdfGoHTML records, in the sales and scope labels

    What it said
    The SDK was described as effectively closed to new buyers, which vendor staff contradict by stating that approved applications can purchase. Its evaluation output was also presented in a way that could read as a free tier. pdfGoHTML’s account, inspection and local processing labels each asserted more than the vendor documents.
    What it says now
    Sales are case-by-case approval, the watermark is labelled as evaluation output, the component split is preserved, and the current domain is cited rather than an old redirect. pdfGoHTML’s fields are scoped to what the documentation supports.

  473. 7a053ac

    The Master PDF Editor and PDF Architect records, in the processing badge

    What it said
    Both records had their processing notes rewritten to describe a mixed reality while the badge above them still read No server involved. The badge a reader sees first is not the note beneath it: the badge is derived from the edition rows, so rewriting the prose changed the caveat and left the headline claim untouched.
    What it says now
    The edition rows carry the processing split, so the badge itself reads Mixed, depends on version and feature. The same fault was found and fixed on several records at once, which is what turned it from a mistake into a rule.

  474. 7a053ac

    The PDF-XChange Editor record, in the watermark note

    What it said
    A conditional watermark note was written into a field the schema does not define. The schema strips unknown keys, so the sentence never rendered, and the type check never complained because an unrecognised key in a data file is not a type error.
    What it says now
    The note is in a field that exists and renders. The larger point is recorded with it: a field that silently disappears is indistinguishable, from the outside, from a field that was never written, and only reading the built page tells them apart.

  475. 7a053ac

    The pdfToolbox record, in the deployment and licensing fields

    What it said
    It carried a categorical no server claim and described the packaging as the documentation does not. The Server and CLI package includes Desktop for profile configuration, with Server using the CLI in the background and Linux shipped under that package, so the boundary between what runs where was drawn in the wrong place.
    What it says now
    The documented packaging, with the categorical claim replaced by a description of self-managed processing.

  476. 7a053ac

    The pdfaPilot record, in the platforms

    What it said
    The installation platforms were incomplete, so a reader on Linux would not have known it was supported.
    What it says now
    Windows, macOS and Linux installation is recorded, including the systemd guidance the vendor publishes.

  477. 7a053ac

    The ABBYY FineReader PDF record, in the evaluation routes

    What it said
    One evaluation description stood for every edition. The vendor’s terms differ by edition: seven days and 100 pages for individual use, and thirty days, five licences and 500 pages for business. The record also blurred Mac creation with Windows editing.
    What it says now
    Each edition’s trial terms are recorded separately, and the platform difference between creating and editing is stated.

  478. 7a053ac

    The Power PDF record, in the editions and their features

    What it said
    Editions and features were flattened together, so installed Standard, Advanced and Mac read as though they were the same as the cloud, Teams and AI services, and perpetual licensing was not scoped to the editions that actually offer it.
    What it says now
    An edition and feature matrix that separates the installed products from the cloud services, with perpetual licensing scoped to retail.

  479. 7a053ac

    The LogrunLab PDF Compare record, in the test claim and the comparison description

    What it said
    A recorded network test was presented as though it established more than it observed, and the record claimed to be the unique place in this index for comparing documents, which is a claim about our own index rather than about the tool.
    What it says now
    The test is framed as what it observed, with its trace, fixtures, browser and date preserved before its request counts are carried forward, and the exclusivity claim is replaced with the truthful statement that this is one place to compare documents rather than the only one.

  480. 7a053ac

    The No Fuss PDF record, in the test narrative

    What it said
    The test was written up as a narrative rather than as a result, so a reader looking for the outcome had to read the account of arriving at it.
    What it says now
    A short result with a link to the evidence behind it.

  481. 7a053ac

    The PrizmDoc record, in the processing and the pricing

    What it said
    Three things ran together: customer-hosted server processing and the vendor’s own cloud, the 300 free cloud transactions described as an ongoing free tier when they are an evaluation allowance, and a pricing note written for our own accounting rather than for a reader.
    What it says now
    The two processing paths are separated, the free allowance is marked as an evaluation, and the pricing is in visitor-facing language.

  482. 7a053ac

    The Apryse record, in the price, the evaluation and the SDK scope

    What it said
    It said no price unit is stated, while the vendor’s pricing guide lists entry packages from 1,500 dollars a year. The seven day keyless evaluation was not distinguished from the demo key that tracks nothing. And local processing was claimed for the product rather than for the configured client-side viewer.
    What it says now
    The entry price with the rest marked as needing a quote, the two evaluation routes kept apart, local processing restricted to the configured client-side viewer, and the event metering and standalone application claims marked as needing contract verification rather than repeated.

  483. 7a053ac

    The Adobe Acrobat record, in the file size and the feature scope

    What it said
    A universal 100 MB limit flattened limits that differ by task, and redaction and accessibility capabilities that belong to the paid desktop editions were attributed to the product as a whole, including the free reader and the online tools.
    What it says now
    Task-specific limits as the vendor’s own FAQ draws them, for ordinary tools, Word and PowerPoint conversion, compression and splitting, with merging counted separately. The free reader, the online tools and the paid desktop editions are separated, and applied removal, hidden data sanitisation and saving are recorded as distinct workflow stages.

  484. 7a053ac

    The empty state on the index, and its show all button

    What it said
    The button cleared the search, task, cost, processing and kind filters but left the saved-only view pressed, so a reader who clicked show all tools was shown one tool.
    What it says now
    It clears the saved-only state and its accessibility value too, with the saved records themselves left alone.

  485. 7a053ac

    The iLovePDF product page title

    What it said
    The page title was cut off mid-description. The field added to hold a proper title was never read by the code that builds titles, which is why the truncated version survived every pass since it was introduced.
    What it says now
    The title reads iLovePDF: PDF tools, features and limitations, and records without one still get a generated fallback.

  486. 7a053ac

    The LlamaParse record, in the pricing

    What it said
    The wording implied a quote is needed, when the vendor publishes a free plan and Starter and Pro figures.
    What it says now
    The published plan figures, with credits explained as the billing unit rather than pages, so a reader comparing it with a per page tool is comparing the right things.

  487. 7a053ac

    The SignWell record, in the accounts, allowances and the edit category

    What it said
    Senders and account free recipients were run together as though they were one kind of user, the web allowance was not separated from the API allowance, and Edit was presented as editing when the vendor labels it form preparation.
    What it says now
    Senders are separated from account free recipients, the two allowances are stated apart, and Edit is named for what it does.

  488. 7a053ac

    The SignNow record, in the batch capability

    What it said
    Batch processing was recorded as not verified, which told a reader nothing while the vendor documents bulk signature invites.
    What it says now
    The documented bulk invites are recorded, with the conflict between the vendor’s own sources kept visible rather than resolved in our favour.

  489. 7a053ac

    The Dropbox Sign record, in the regional storage note

    What it said
    Multiple regions was published as though where a file is stored also decided where it is processed. It does not: the United States is the default region at rest and premium tiers can choose others, but active editing, sending and signing temporarily processes copies on United States servers even then.
    What it says now
    The scoped explanation, which is the distinction a reader with a residency requirement actually needs.

  490. 7a053ac

    The DocuSign record, in the free sending and regional statements

    What it said
    It said there is no free way to send, and the regional detail was stated as one thing rather than several.
    What it says now
    Sending is available during a free trial with ongoing sending requiring a paid plan, and the regional note separates data at rest from transaction data and from integrations that may cross regions.

  491. 7a053ac

    The PandaDoc record, in the storage claim and the free plan

    What it said
    Stored for the life of the account was published as fact, and the free plan’s limits were not distinguished from its API eligibility.
    What it says now
    The storage statement is marked unverified, the five template limit is recorded, and the free plan’s API eligibility is separated from the limits of the plan itself.

  492. 7a053ac

    The OpenSign record, in the durability warning

    What it said
    It warned that the default database clears when the service restarts. The current compose file mounts a data volume and a named document volume, so the warning contradicted the configuration the project ships.
    What it says now
    What the current compose file does is recorded, described as a reading of the configuration rather than as a durability test, because a mounted volume is not proof that data survives an upgrade.

  493. 7a053ac

    The PDF Candy record, in the file size limit

    What it said
    The 500 MB maximum was published as a general allowance.
    What it says now
    It is recorded as a paid plan allowance, which is the difference between a limit and a reason to pay.

  494. 7a053ac

    The PDF2Go record, in the batch limits

    What it said
    Current pricing and older help pages were merged into one set of limits, so a reader got numbers that had been superseded without being told.
    What it says now
    The current paid batch support is recorded, up to 200 files and 64 GB per task, beside the help pages still describing 400 files and 8 GB, recorded as a conflict rather than averaged. Desktop offline is a vendor claim rather than a tested guarantee.

  495. 7a053ac

    The AvePDF record, in the retention and file size

    What it said
    It said files are deleted about thirty minutes later, which conflates how long a file is reachable with when it is removed.
    What it says now
    Files are accessible for up to 30 minutes, the free service’s separate retention permission is kept before any outbound action, and the documented 128 MB maximum is scoped to the tools that state it rather than applied to all of them.

  496. 7a053ac

    The PDFescape record, in the limits

    What it said
    One set of limits stood for every edition.
    What it says now
    Per edition: free online at 10 MB and 100 pages, Premium and Ultimate at 40 MB and 1,000 pages, with history of seven days, thirty days and no expiration respectively.

  497. 7a053ac

    The DocHub record, in the redaction verdict

    What it said
    It stated that redacted content is definitively not removed, which is a verdict resting on one of two vendor sources that disagree: the help article describes a movable whiteout while the pricing matrix claims permanent removal.
    What it says now
    Behaviour not independently tested, with the conflict stated. An exported file would have to be inspected before any verdict, and that has not been done.

  498. 7a053ac

    The Notability record, in the free plan and the pricing

    What it said
    Current pricing and older subscription guidance were combined into a plan that does not exist.
    What it says now
    The current free notes statement with its date, the March 2026 help page conflict recorded rather than merged, and annotation and markup added as a capability.

  499. 7a053ac

    The Master PDF Editor record, in the processing note

    What it said
    It said no server is involved and that the only network step is activation, which the activation guide contradicts: update checks are documented.
    What it says now
    Desktop processing, with the statement that network behaviour has not been independently tested rather than a claim about what the application never does.

  500. 7a053ac

    The PDF Studio record, in the trial and licensing

    What it said
    The watermarked free demo and the full access free trial were collapsed into one unknown trial watermark field, though the conditions differ and one of them requires a credit card.
    What it says now
    The two are stated separately, and subscriptions are recorded alongside perpetual licensing rather than as alternatives to it.

  501. 7a053ac

    The PDF-XChange Editor record, in the watermark and the AI note

    What it said
    The free watermark was stated unconditionally, and AI features were presented without saying where the data goes.
    What it says now
    The free watermark state is conditional, optional AI features send data to configured providers, and the batch tooling documentation is kept separate from the editor rather than merged into it.

  502. 7a053ac

    The Soda PDF record, in the free tier and the price

    What it said
    The free tier was a flat No, and the converter help page’s condition of two files a day or 3 MB was turned into a rule rather than quoted.
    What it says now
    Limited free online tools with the per tool exceptions named, the ambiguous condition quoted as written, and the price recorded as not established in this review rather than guessed.

  503. 7a053ac

    The PDF Architect record, in the processing note

    What it said
    It said no server is involved, which the vendor’s own statement does not support: core documents stay local by default, with activation and electronic signature workflows contacting services.
    What it says now
    The vendor’s statement that core documents stay local by default, the activation and signing paths named, and European residency marked unverified for those workflows.

  504. 7a053ac

    The PDFelement record, in the conversion limits

    What it said
    The record gave one figure for conversion pages while the vendor’s own Windows FAQ gives three in its table and five in its prose.
    What it says now
    The conflict is recorded rather than resolved, so a reader planning around the limit knows the vendor disagrees with itself.

  505. 7a053ac

    The Nitro PDF Pro record, in the processing badge and the API row

    What it said
    The badge implied local processing while Smart Redact, extraction and PDF chat are online services, and an HTTP API row appeared on a record for a desktop application.
    What it says now
    The badge reads Mixed, depends on version and feature, the Smart Tools are identified as online services, and the API row is removed from a record that is about the desktop product.

  506. 7a053ac

    The pdf-redactor record, in the removal claim beside the badge

    What it said
    The removal badge stood alone, so a reader could take it as complete sanitisation. What the tool does is narrower: it substitutes the text layer and selected metadata.
    What it says now
    The scope is repeated beside the badge, where the claim is made rather than further down the page, and complete sanitisation is not implied.

  507. 7a053ac

    The redaction guide, in the quick check it recommends

    What it said
    It offered the copy and search check as a certainty, which it is not. Finding the text again means the redaction failed, and that is the whole of what it establishes. Finding nothing establishes nothing about images, metadata, attachments, revision history, hidden objects or unmapped fonts.
    What it says now
    The check is described as a failure detector, with both halves stated: a hit is proof of failure, and a miss is not proof of success. The guide no longer promises a certainty it cannot deliver.

  508. 7a053ac

    The redaction guide, in its counts of exceptions and entries

    What it said
    The page carried hardcoded counts, eight exceptions, three entries, all but one, which had already drifted once and would have drifted again the next time a record changed.
    What it says now
    Every count is derived from the evidence the records hold, so the page cannot disagree with the data behind it.

  509. 7a053ac

    The accessibility guide, in the law and the standard

    What it said
    Section 508 and EN 301 549 were both presented as procurement law. Section 508 is the law that obliges federal agencies, and EN 301 549 is the standard those purchases are measured against, so the guide had the relationship backwards.
    What it says now
    The law and the standard are stated as what each one is.

  510. 7a053ac

    The accessibility guide, in what authoring is said to achieve

    What it said
    It said that authoring means the problem is never created, which overstates what good authoring does and quietly excuses the person producing the document.
    What it says now
    Authoring catches structure problems before a finished document rather than repairing them afterwards, and the author still answers the accessibility question. A PDF/UA pass is also recorded as not settling every applicable requirement.

  511. 7a053ac

    The local processing guide, in its central heading

    What it said
    The heading read No server in the path at all. An architectural assessment cannot establish that, because installed software and embedded libraries can use cloud features, remote resources, or a server the reader runs.
    What it says now
    It reads No vendor service in the path, which is the claim the evidence actually supports, and the difference is one a reader can act on.

  512. 7a053ac

    The method page, in the figures and their complements

    What it said
    Two complements were published, that everything else sends it to a server and that the rest draw over it, and the figures themselves carried no indication of how they were arrived at. A total mixing a test with a reading with a vendor’s wording is not one kind of evidence.
    What it says now
    The complements are removed and every figure carries its basis: an architectural assessment, vendor documentation, or something read from the vendor’s own pages.

  513. 7a053ac

    The open-source licensing guide, in what each licence family requires

    What it said
    Two things. It said a permissive licence asks for attribution and nothing else, while Apache 2.0 also asks you to ship a copy of the licence, to state that you changed files, and to carry any NOTICE file with the distribution. And it said a copyleft licence asks you to publish the source of what you distribute, which flattened the differences between the LGPL, the MPL and the GPL, and described the AGPL as though everything you run becomes public.
    What it says now
    Each licence’s obligations are stated, the AGPL is described as a source offer to the people using the service over a network, and the guide no longer promises a simplicity the licences do not have.

  514. 7a053ac

    The open-source licensing guide, in how records are grouped by licence

    What it said
    The grouping took the first licence name in the field and stopped. A record reading MPL-2.0 for the core with a bundled Ghostscript under AGPLv3 was therefore filed under MPL alone, and its AGPL dependency disappeared from the page entirely. A reader asking whether they could ship it was never shown the licence that would decide the answer.
    What it says now
    Every licence the field names is found and a record is filed under the most restrictive one, because a reader deciding whether they can ship something with an AGPL dependency is asking an AGPL question. Records naming more than one licence are marked so that a single word beside the name does not stand for the whole position.

  515. 7a053ac

    The PDFgear record, in the processing verdict

    What it said
    The badge said it stays on your device while the note directly underneath said certain advanced conversions run on the vendor’s cloud servers and that the AI copilot is cloud based. It also scoped itself to the web service, when the vendor’s local processing claim is about the installed builds, so the evidence and the scope were the wrong way round.
    What it says now
    A mixed verdict that names both halves: the Windows, macOS, iOS and Android builds read and write local files, and some conversions and the AI copilot are cloud features inside those builds. A mixed verdict is no longer also allowed to say it covers one service, because that sentence contradicted the verdict above it, and six task pages now say that one tool here processes some versions on your device and runs others on a server.

  516. 7a053ac

    The footer on every product page

    What it said
    It was a binary: either reviewed by a named person, or Compiled from published sources. Not yet reviewed by a person. No record carries a reviewer, so all 153 printed the second, including 147 that carry dated checks describing work somebody had actually done. The line read as though nothing had been checked and contradicted the page above it. The heading over that section said Verified details, above a table mixing vendor statements, our own readings and fields nobody had established. And two disclosures printed on records they did not apply to: the hash sentence on records with no claims, and the currency conversion note on free and open source records whose unit price is deliberately never converted.
    What it says now
    Four separate states, each saying what is true of that particular record: whether an editor has signed it off, which is explicitly not the same as nothing having been checked; how many claims came from the vendor’s own pages and the dates they were read between; when the vendor’s site was last checked; and whether anything was run by us. The heading reads Details, and where each comes from, the hash sentence appears only where there are claims to hash, and the currency note only where a figure was actually converted.

  517. 7a053ac

    The edit and sign tags, and the two task pages they appear on

    What it said
    Apple Preview carried an edit tag while its own note said Apple states you cannot edit PDF text in Preview. Okular carried edit while saying it annotates rather than edits. Microsoft Edge carried sign while saying it cannot create digital signatures, only view and validate them. The job’s own description promised all of that, and a record had no way to say which part of a job it does.
    What it says now
    A job entry can carry a scope, used only where the record’s own description is narrower than the job name, so it marks exceptions rather than inventing a second taxonomy. Preview reads annotate, not existing text. Edge reads view and check a signature, not create one. The task pages say how many tools do less than the name of the job suggests, and the merge page carries no note, correctly, because none of its tools is narrower.

  518. 7a053ac

    The Edge and Chrome records, in the processing verdict and the scoping sentence

    What it said
    Both carried a finding that no server is in the path, because the browser’s own viewer processes locally, and both were then printed with This covers the web service. We have not checked desktop app processing. The one thing a finding of no server in the path cannot be true alongside is a sentence about which service it covers.
    What it says now
    The scoping sentence is suppressed for that verdict as well as for the mixed and self-hosted ones, so the contradiction is gone and the 20 profiles still carrying it are the ones whose finding really is about a service. Both records name the optional cloud action separately: Edge’s protected file viewing needs a signed in identity and a licence, and Chrome’s save to Drive uploads the open file. Retrieving a web hosted document is described as retrieval rather than as our file being sent anywhere, so opening one does not turn a browser into a cloud edition.

  519. 7a053ac

    The accessibility task page, and the accessibility tags on six records

    What it said
    The accessibility tag said a tool handled accessibility work without saying which part of it. The records already said which, and the page did not, so a tool that checks, a tool that repairs an already tagged document, and a tool that converts to another format were all filed the same way, and a reader could take a passing check for an accessible document.
    What it says now
    Each record’s accessibility scope is named: the checkers check what a machine can check and do not repair; one repairs an already tagged document but does not tag one; two make another format; and one validates and reports. The task page carries the caveat that a full set of green results means the machine checkable requirements passed, not that the document is accessible, and that the human review is still needed.

  520. 7a053ac

    The Foxit PDF SDK record, in the trial watermark

    What it said
    The trial watermark was left as an unknown rather than recorded as what the vendor documents.
    What it says now
    It is recorded as vendor documented and scoped to the thirty day desktop and server trial, so a reader knows which trial they would actually be getting rather than that there is one.

  521. 7a053ac

    The No Fuss PDF listing

    What it said
    It said that filling places an image of your answer over the page rather than changing the document, and that the tool was therefore the wrong choice if the values needed to be machine readable afterwards. The first half was an assumption about the mechanism dressed as an observation, and the second half was false: an answer a reader types is selectable text and can be extracted from the finished file. The same listing also described the tool as working on the scanned and flattened forms that a text based tool has to turn away, which asserts what competing tools do and was never tested here.
    What it says now
    That filling adds a new text layer over the page, so the answer is selectable and extractable while the words already printed in a scan stay image only, and that a reader who needs the scanned words themselves to be searchable needs OCR. The line about what other tools do is gone. The measurement is named on the listing rather than implied, and it was re-run on our own document rather than left resting on the example the operator supplied: an image only form built here was filled, the answer came back out as selectable text, and the scanned words did not.

  522. 7a053ac

    The IronPDF profile, in the licence field and the worth-knowing note

    What it said
    It said redistribution and SaaS use require Royalty-Free Redistribution Coverage as a paid add-on, that every tier from Lite up could be deployed provided you bought it, and that the coverage costs US$1,999. The vendor says the opposite on its own licensing page: "OEM Redistribution & Air-gapped dev environment - Not available for the Lite license. Please choose Plus, Professional, or Unlimited for coverage options."
    What it says now
    Both passages quote that exclusion and say the cheapest route to redistribution starts one tier up rather than at US$999 plus an add-on.

  523. 7a053ac

    The price row on every product page

    What it said
    A figure converted from another currency was printed as a bare price with nothing saying so. IronPDF charges US$999 and its page read "From £740.96 one-off", with the exchange rate mentioned only inside a long sources list further up. The schema has no field recording whether a vendor prices in GBP or whether the number was converted, so a reader taking the rail at face value saw a number the vendor does not charge and which moves with the rate.
    What it says now
    Every product page says prices are shown in GBP, that some vendors price in another currency and those figures are converted at the European Central Bank reference rate for the date the page was read, and that the figure deciding what you actually pay is the vendor’s own, on the page linked above.

  524. 7a053ac

    The pricing page, the homepage, the submit page, and the privacy policy

    What it said
    The Featured tier promised a "Monthly click report" and two other pages listed "click reporting" as something a free listing gets, while nothing reported clicks to anyone. Separately, the privacy policy described a public "site checker" in about a dozen places, including what it would collect and why, and no such checker existed.
    What it says now
    The tier says clicks are counted per product per day and that the monthly report is not automated yet, the other two pages say "click counts", and the privacy policy states before any of it that the site checker is not available and that none of the described data has been gathered from anyone.

  525. 7a053ac

    The evidence drawer on the iLovePDF profile, under a claim we refuted

    What it said
    It said the recorded hash was not reproduced and explained that the page’s pricing markup had changed, ending "the claim is about the redaction tool and is unaffected". So the only substantive thing in the drawer was about a price string, and the word “unaffected” sat directly beneath a claim our own test had refuted.
    What it says now
    It says the pricing change has nothing to do with the claim, names the refutation and what the test found, and explains why it is the only drift note on the site that does not end "the verdict stands".

  526. 7a053ac

    The homepage hero

    What it said
    It said that every vendor claim carries the date and the SHA-256 of the page we read. All 294 claims cite a vendor page, but only 215 carry a hash, so the sentence was wrong about the coverage on the most-read page on the site.
    What it says now
    It says every vendor claim is quoted from the vendor’s own page and that, where we kept the evidence, the claim also carries the date and the SHA-256 of the text we read. The launch tagline and four other pieces of launch copy were corrected the same way.

  527. 7a053ac

    The launch announcement for this site

    What it said
    It said every claim carries the SHA-256 of the page it was read from. It also said the site can tell when a vendor changes their own words, and that there was a backlog of 148 open items.
    What it says now
    It says the claim carries the hash where we hold the evidence and that the coverage is counted on every build; that the site notices when a source comes back different, which is not the same thing; and that there is a public backlog, without a figure that goes stale.

  528. 7a053ac

    The PrizmDoc profile

    What it said
    It stated that the pages read describe neither retention nor residency, that the vendor had published nothing about where documents are processed or for how long.
    What it says now
    It quotes Accusoft's own hosting FAQ, which names AWS us-east-1 in Virginia, and its server documentation, which says work files are deleted 24 hours after creation by default, with the scope limits named.

  529. 7a053ac

    The AvePDF profile

    What it said
    A published finding said the redaction engine is not AvePDF’s own, which contradicted the same record’s own notes.
    What it says now
    It says the engine is the group’s own commercial technology, quoting the vendor: "We develop all the AvePDF engines" and "AvePDF is built on Nutrient’s enterprise-grade PDF SDK".

  530. 7a053ac

    The CamScanner profile

    What it said
    It described the developer offering as an agent platform and a REST API.
    What it says now
    It describes document capabilities delivered over MCP, and notes that the vendor’s own documentation says "This is not a new agent platform and does not replace the agent you already use". The page it had been sourced to returns a 404 in a real browser.

  531. 7a053ac

    The Poppler profile

    What it said
    It said the licence is GPL-2.0-only, taken from distribution packaging, and that upstream COPYING could not be read.
    What it says now
    It says GPL-2.0-or-later for Poppler’s own changes, over an xpdf base under GPL-2.0, quoting the project’s source headers: "licensed under GPL version 2 or later". Upstream reads fine now, and the artefacts are held.

  532. 7a053ac

    The downloadable dataset, and the page describing it

    What it said
    The dataset published Stirling PDF as untested while its profile said we had watched it keep the file on our own instance. The page beside it said three tools had been canary-tested when eight records said they had.
    What it says now
    The export is built from the same label map the pages use, so the two cannot disagree again, and the canary count is computed from the records rather than typed in.

  533. 7a053ac

    The privacy and trust page

    What it said
    It said we record "the product, the day, and a count", while the stored row also carried a full timestamp, and it said the derived value is deleted once the day is over when detail rows outlive the day.
    What it says now
    It says the day, the time and a count, and that detail rows are deleted within two days. The dedupe sentence was corrected too: it is once per visitor per product per day, not once per visitor per day.

  534. 7a053ac

    The SnackPDF versus SimplePDF comparison

    What it said
    The prose said SimplePDF’s whiteout covers text while SnackPDF "has no redaction feature at all". Both clauses were backwards, and the comparison table directly above the sentence rendered both correctly.
    What it says now
    It says SnackPDF removes the content and that the verdict came from a test rather than a reading, while SimplePDF has no redaction feature.

  535. 7a053ac

    The Camelot profile

    What it said
    Three claims carried history entries saying the vendor’s page no longer carries the sentence we quoted.
    What it says now
    Each carries a correction noting that the page never changed: our comparison tokenised Markdown into the claim and could never match it. The artefact and its hash are unchanged, because the read did not change, only the reading of it.

  536. 7a053ac

    The CommonLook PDF profile

    What it said
    It called the desktop edition a "desktop download", which left a reader looking for a plug-in with no way to tell what they were getting.
    What it says now
    It says there are two editions and that the second is a plugin to Adobe Acrobat Pro or Pro DC, quoting Allyant’s own support documentation.

  537. 7a053ac

    The alternatives and comparison indexes, and one alternatives page

    What it said
    They promised to show "what you pay month to month" above tables rendering weekly, one-off and annual figures.
    What it says now
    They say what each tool costs to start, with the period that figure covers. Of the 57 records carrying a price, 26 are monthly, 16 annual, 12 one-off and 3 weekly.

  538. 7a053ac

    The pricing page and the homepage FAQ

    What it said
    The £249 tier included "and the newsletter", which does not exist, and the FAQ promised that a profile "links to its check history", which no profile did.
    What it says now
    The tier lists a job page position and the hero. The FAQ says the profile shows the latest check and when it was taken.

  539. 7a053ac

    The accessibility guide and the Goodnotes profile

    What it said
    The guide said two of its four validating tools are Acrobat plug-ins, when one is and the second it meant was the non-validating fifth. The profile said some capabilities on the vendor’s site are labelled as coming later.
    What it says now
    The guide says one of those four, and adds the count a reader wants: two of the five need Acrobat installed. The profile drops the roadmap label, which was searched for on two surfaces and is not there.

  540. 7a053ac

    The PDF Architect profile, and the redaction guide

    What it said
    The record put redaction in a "Secure module" and said the vendor’s own pages disagreed about which module held it. The guide repeated the same claim in hand-written prose.
    What it says now
    Both say redaction is included from Pro upwards, quoting the pricing page, with no separate Secure module. The guide line survived a record-level fix, which could not reach it.

  541. 7a053ac

    The ABBYY FineReader and ReportLab profiles

    What it said
    Both carried an instruction not to publish a claim, which a later passage in the same record had already answered.
    What it says now
    Both instructions are retired with a note. ABBYY’s redaction claim rests on the vendor’s own sentence, and ReportLab publishes no CLI claim because the project declares no console entry point.

  542. 7a053ac

    The Microsoft Lens profile

    What it said
    It had been published with an instruction to verify that the discontinued banner renders, which had not been done.
    What it says now
    The banner is verified on the built page and on the live profile, and the note records the date. The retirement is the useful information on that page, so the banner is the whole record.

  543. 7a053ac

    The pdf-redactor and DocHub profiles

    What it said
    One asked which project the site intended before citing an author. The other asked for a human eye on the vendor’s "encrypted end-to-end" wording.
    What it says now
    The first names the authors the project itself names in its README, and records that the Free Press wrapper everyone worried about is an archived fork of a different tool. The second keeps the vendor’s words and leaves the existing finding, which had already explained that "end-to-end" rests on transport encryption only.

  544. 7a053ac

    The Adobe PDF Extract API, Azure Document Intelligence and PDFIX profiles

    What it said
    One recorded a free-tier contradiction that was stale metadata in a meta tag rather than a competing offer. One carried a caveat that its own figures come from a different Microsoft endpoint. One flagged a euro figure against a dollar figure as unreconciled.
    What it says now
    The free tier is 500 transactions a month, in the visible body. The Azure figures are re-verified and the distinction between converted estimates and list prices is named. PDFIX prints one seat-year in two currencies, and its Enterprise tier exists only inside the FAQ.

  545. 7a053ac

    The build check that guards every hash

    What it said
    It failed on every claim naming an evidence file whenever the evidence archive was absent, which is always in continuous integration. So the check could never pass there and the deploy was red on every push.
    What it says now
    It enforces format and reports coverage when the archive is absent, says plainly that bytes were not checked, and still fails on a genuinely missing file when the archive is present.

  546. 7a053ac

    Our own listing, and the rule for naming people

    What it said
    Our own product briefly carried a first name in its "Operated by" line, after the idea of crediting makers was floated.
    What it says now
    The field is an entity and nothing else, so it is null, and the schema records why: the plausible sources for a maker’s name are a handle, a commit email or a whois record, and a wrong attribution on an indexed page is worse than no name.

Not on this list

Vendor pages that have moved since we read them are not corrections, and they are not here. Each claim carries a Standing line saying whether the recorded hash still matches, and the honest limit of that check is that the hash covers the whole page rather than the sentence, so a commit counter moving looks the same as a rewording. That is explained in full on the method page.

If something on this site is wrong and is not listed here, that is worth more to us than a new listing. Write to calum@dialegein.com, or use the corrections route on the trust page. If it turns out to be wrong, it gets an entry here with the commit that fixed it.