Loading page
About this tool
Inspect PDF and modern Office documents locally for observable title, language, structure, table, and image-description accessibility cues.
The Accessible Document Checker opens a PDF, DOCX, XLSX or PPTX in the browser and reports the accessibility markers actually present in the file: a document title, a default language, heading or outline structure, table header rows, alternative-text metadata on images, and bookmarks. For PDFs it scans the raw bytes for /StructTreeRoot with /Marked true, /Lang, /Title, /Alt entries against image objects and /Outlines; for Office files it unzips the package and reads docProps/core.xml and the document XML for language, heading styles, table header markers and drawing descriptions. Each result comes back as present, needs review, or manual — it never claims conformance it cannot observe.
Open Accessible Document Checker on AltFTool — it loads instantly in your browser.
Press the 'Choose PDF, DOCX, XLSX, or PPTX' picker under 'Document file' and select a file up to the stated maximum of 20 MB.
Press 'Inspect locally' to parse the file in the page; 'Clear' drops it again without anything being uploaded.
Read the screening summary — Marker present, Needs review, Manual only — then press 'Export counts-only JSON' to save document-accessibility-cues-counts-only.json.
PDF, Word, Excel and PowerPoint are inspected with format-appropriate markers instead of a single generic file scan.
Findings are present, review or manual, so items a static scan genuinely cannot judge — reading order, decorative intent — are marked as needing a human rather than silently passed.
The file is parsed in the page under explicit size limits and never uploaded, which matters for unpublished or confidential documents.
PDF, DOCX, XLSX and PPTX, up to 20 MB per file. Office packages are additionally bounded at 3,000 entries and 80 MB uncompressed, with each XML part read to 2 MB, so a very large deck or workbook is inspected partially and flagged as truncated rather than parsed unsafely.
No. It means the structural markers it looks for were found. Reading order, heading hierarchy, whether alt text is actually meaningful, table semantics, colour contrast and form labelling all need a human and assistive technology. The checker marks those as manual precisely so they are not mistaken for passes.
Because PDF markers can sit inside compressed object streams that a raw byte scan cannot see, so absence of a marker is not proof of absence of the feature. An encrypted PDF has the same effect. Confirm in a PDF editor's accessibility checker before concluding the tags are missing.
No. Office files are unzipped with a JavaScript ZIP reader and PDFs are decoded as bytes, all inside your browser tab. Nothing is transmitted, so you can check draft contracts, unreleased reports or student records without them leaving the machine.