Document is not tagged
The PDF has no logical structure tree, so assistive technology cannot navigate the content semantically.
From raw PDFs to well-done.
Checks
The current PDF accessibility check library ChefPDF is being built to evaluate.
214 planned checks
The checks below are scaffolded as a deterministic PDF/UA-oriented library. They do not run real PDF parsing yet, but each item has a stable ID for future implementation, reporting, and remediation work.
No checks match the current filters.
Document Structure
14 checksMarked property is missing or false
The catalog does not identify the document as tagged even when structural tags may be present.
Structure tree root is missing
The StructTreeRoot entry is absent or invalid, preventing a reliable logical reading model.
Parent tree is missing or inconsistent
Marked content references cannot be resolved back to the structure tree through a valid parent tree.
Marked content IDs are duplicated
MCID values are reused in a way that makes content-to-tag associations ambiguous.
Tagged content is not referenced by the structure tree
Visible or meaningful content exists in marked content sequences that are not connected to structural elements.
Structure tree references missing page content
Tags point to content items that cannot be found on the referenced page.
Page content is untagged
Visible content appears on a page without marked content identifiers or structural ownership.
Content is tagged as artifact incorrectly
Meaningful text, images, or controls are hidden from assistive technology as artifacts.
Decorative content is not artifacted
Headers, footers, page numbers, backgrounds, borders, or purely decorative marks are exposed as meaningful content.
Role map contains invalid mappings
Custom structure types are not mapped to valid standard structure types.
Namespaces are missing for custom structure types
Custom tags are used without enough namespace information to interpret them reliably.
Page object lacks structural parent information
Page dictionaries do not expose the structural parent links needed to resolve tagged content.
Incremental updates leave stale tags
The PDF contains old structure references from prior edits that no longer match current page content.
Tags and Semantics
14 checksSemantic tag does not match visual purpose
Content is tagged with a structure type that misrepresents its actual role.
Paragraph text is tagged as generic containers
Meaningful body text is exposed through Div, Span, or other generic tags instead of paragraph-level semantics.
Document title is not represented structurally
The visible document title is missing or not exposed as an appropriate heading or title structure.
Quotes are not tagged semantically
Block quotations or inline quotations are tagged as plain paragraphs or spans.
Code samples are not tagged semantically
Program code, commands, or literal text are not exposed with a suitable code-related structure.
Footnotes are not associated with references
Footnote content is not linked to the corresponding reference in the reading order.
Endnotes are not associated with references
Endnote content and its invocation are not structurally connected.
Captions are not associated with figures or tables
Caption text is detached from the figure, chart, or table it describes.
References are not tagged as references
Bibliographic references, citations, or cross-references are exposed as plain text without reference semantics.
Non-structural layout tables are tagged as data tables
Layout-only table structures are exposed as semantic tables, adding noisy navigation for assistive technology.
Semantic content is split across unnecessary spans
Text is fragmented into excessive inline tags that interrupt meaningful reading and selection.
Whitespace is represented as meaningful content
Spacing characters, empty paragraphs, or layout-only text runs are exposed as content.
Abbreviations lack expansion where needed
Abbreviations or acronyms are not expanded when their meaning is not clear from context.
Replacement text is missing for symbolic content
Symbols, ligatures, or custom glyphs do not provide ActualText or equivalent semantic text.
Reading Order and Reflow
8 checksReading order does not match intended meaning
The logical structure order differs from the sequence needed to understand the document.
Multi-column content reads across columns
Columnar layouts are ordered horizontally across columns instead of within each column sequence.
Sidebar content interrupts main flow
Pull quotes, sidebars, marginalia, or callouts are inserted in a way that breaks the primary reading flow.
Repeated headers or footers interrupt reading
Running headers, footers, or repeated navigation content are exposed repeatedly in the logical flow.
Hyphenated words are exposed incorrectly
Line-break hyphenation causes words to be read, copied, or searched incorrectly.
Text extraction order differs from tag order
The content stream order and the structure tree order conflict in ways that can affect assistive technology.
Reflow loses meaningful relationships
When content is reflowed, labels, values, captions, or grouped items are separated from each other.
Hidden or clipped text is exposed
Invisible, clipped, or off-page content appears in the accessible reading stream.
Headings
6 checksHeading levels skip unexpectedly
Heading hierarchy jumps over levels without a meaningful structural reason.
Visual heading is not tagged as heading
Text styled as a heading is exposed as a paragraph or generic tag.
Non-heading text is tagged as heading
Large, bold, or decorative text is incorrectly exposed as a heading.
Multiple top-level headings conflict
The document exposes multiple top-level headings where the hierarchy suggests a single title should lead the structure.
Heading text is empty
A heading tag has no meaningful text or accessible replacement.
Heading order differs from visual section order
Headings are structurally ordered in a way that does not match the document sections users see.
Lists
6 checksList is not tagged as a list
Visually grouped list items are exposed as plain paragraphs or lines.
List item labels are missing
Bullets, numbers, or item markers are not represented in the list item structure.
List item body is missing
List items do not contain a valid body element for the item content.
Nested list structure is flattened
Nested lists are exposed as a single flat list, losing hierarchy.
List numbering is incorrect
Ordered list labels do not match the visible numbering or sequence.
Pseudo-lists are not grouped
Repeated label-value or step patterns are not grouped in a navigable list-like structure.
Tables
12 checksData table is not tagged as a table
Rows and columns are visually present but missing table structure semantics.
Table headers are missing
Header cells are not identified for a data table that requires row or column headers.
Header scope is missing
Header cells do not define whether they apply to rows, columns, or both.
Header associations are missing for complex table
Complex table data cells cannot be programmatically associated with all relevant headers.
Table row structure is invalid
Table cells are not contained in valid table row structures.
Table cell spans are incorrect
RowSpan or ColSpan values do not match the visual table grid.
Empty table cells lack intent
Blank cells are not identified as intentionally empty or are used only for layout spacing.
Table caption is missing or detached
A visible table title or caption is not associated with the table structure.
Table summary is missing where needed
A complex table lacks explanatory text needed to understand its organization.
Split table continuity is not conveyed
A table split across pages or sections does not preserve header and continuation relationships.
Merged visual cells are structurally separate
Cells that visually span a region are represented as multiple unrelated cells.
Data table uses presentation-only artifacts
Lines, shading, or positioning carry meaning that is not represented in the table structure.
Images and Graphics
10 checksMeaningful image lacks alternative text
An image, figure, or graphic that conveys content has no accessible description.
Decorative image has noisy alternative text
Purely decorative imagery is announced instead of being artifacted.
Alternative text is too vague
Image descriptions such as image, photo, or graphic do not convey the image purpose.
Alternative text repeats adjacent text
The figure description duplicates nearby caption or body text without adding useful meaning.
Chart lacks data-equivalent description
A chart, graph, or visualization lacks a text equivalent for trends, values, and conclusions.
Diagram lacks structural explanation
A process, flow, map, or technical diagram is not described in a meaningful sequence.
Image of text has no text alternative
Text rendered inside an image is not available as real text or equivalent replacement text.
Icon-only meaning is not described
Icons that communicate status, action, or category lack an accessible name or explanation.
Watermark is exposed as content
Watermarks or background marks are announced as part of the document reading flow.
OCR text conflicts with image text
Recognized text over an image is inaccurate, duplicated, or ordered differently from the visible text.
Links and Annotations
8 checksLink lacks an accessible name
A link annotation does not expose meaningful link text or alternate description.
Link purpose is ambiguous
Repeated text such as click here or read more does not identify the destination or action.
Link annotation is not nested in Link tag
A clickable annotation exists without a corresponding logical Link structure element.
Link tag has no annotation
A logical Link structure exists but is not connected to an actionable PDF annotation.
Clickable area does not match visible link
The annotation rectangle is too small, too large, or offset from the visible link text.
Link target is broken
The destination, named target, page reference, or external URL cannot be resolved.
Non-link annotation is unlabeled
Comments, popups, stamps, attachments, or other annotations do not expose a meaningful label.
Annotation tab order is illogical
Keyboard traversal through annotations does not follow a meaningful order.
Forms
10 checksForm field lacks accessible name
An interactive field has no programmatic label, tooltip, or alternate name.
Visible label is not associated with field
The label text near a form field is not connected to the field object.
Required field state is not conveyed
A required input is visually indicated but not programmatically exposed.
Field instructions are not associated
Help text, format requirements, or constraints are not connected to the relevant field.
Validation errors are not associated
Error messages are not programmatically connected to the field they describe.
Radio group lacks group name
Related radio buttons do not expose the shared question or group label.
Checkbox meaning is unclear
A checkbox has an ambiguous label or depends only on visual layout for meaning.
Choice field options are unlabeled
Combo box or list box options do not expose clear text labels.
Form tab order is illogical
Keyboard navigation through fields does not match the intended completion order.
Submit or reset button is unlabeled
Button controls do not expose the action they perform.
Language and Text
10 checksDocument language is missing
The PDF does not define a default natural language for assistive technology.
Document language is invalid
The document language value is malformed or not a recognized language tag.
Language changes are not marked
Passages in another language are not tagged with their own language.
Text lacks Unicode mapping
Fonts or glyphs do not map reliably to Unicode, causing incorrect reading, search, or copy behavior.
Ligatures lack ActualText
Ligatures or presentation glyphs do not expose the intended underlying characters.
Math content lacks accessible representation
Equations or formulas are not exposed with usable text, MathML, alternate text, or a clear explanation.
Reading pronunciation is ambiguous
Symbols, units, or specialized notation lack expansion where pronunciation would be unclear.
Scanned text is not OCRed
Image-only pages contain text that is not available to assistive technology or text search.
OCR text is inaccurate
Recognized text contains errors that change meaning or prevent reliable reading.
Text order inside lines is corrupted
Characters or words are stored in an order that reads incorrectly despite appearing visually correct.
Color and Visual Presentation
8 checksColor alone conveys information
Status, grouping, required state, or meaning depends only on color without text or structural support.
Text contrast may be insufficient
Text and background colors may not provide enough contrast for readable presentation.
Non-text contrast may be insufficient
Meaningful icons, chart lines, form boundaries, or controls may not have enough contrast.
Important content is too small
Text or symbols may be visually present but difficult to read due to small size or scaling.
Text is obscured or overlapped
Layering, clipping, transparency, or positioning makes content difficult to read visually or programmatically.
Focus indication is missing for interactive content
Focusable annotations or fields do not provide a visible focus state in common PDF viewers.
Chart legend depends only on color
Chart categories or series cannot be distinguished without color perception.
Low-resolution raster content harms readability
Rasterized text, scans, or graphics are too degraded for reliable visual or OCR interpretation.
Navigation and Metadata
8 checksDocument title metadata is missing
The PDF metadata does not provide a meaningful document title.
Viewer title setting does not use document title
The viewer is not instructed to display the document title instead of the file name.
Bookmarks are missing for long document
A long or complex document lacks bookmarks or outlines for efficient navigation.
Bookmarks do not match heading structure
Document outline entries are missing, mislabeled, or inconsistent with structural headings.
Page labels are missing or misleading
Logical page labels do not match visible page numbering or expected navigation.
Initial view hides relevant content
The document opens at an inappropriate page, zoom, or pane for accessible use.
Metadata language conflicts with document language
Metadata and catalog language values disagree or are inconsistently applied.
Reading direction is missing where needed
Right-to-left, vertical, or mixed-direction content lacks explicit direction information.
Media and Embedded Content
8 checksAudio lacks transcript
Embedded audio content has no text transcript.
Video lacks captions
Embedded video with speech has no synchronized captions.
Video lacks audio description where needed
Important visual-only information in video is not provided through audio description or text alternative.
Embedded file lacks accessible description
Attachments or file annotations do not expose meaningful names, descriptions, or purpose.
3D content lacks accessible alternative
Embedded 3D objects do not provide an equivalent description or fallback.
Scripted behavior blocks accessibility
JavaScript actions interfere with keyboard access, reading, focus, or assistive technology behavior.
Flashing content may trigger seizures
Animated or multimedia content may flash at unsafe frequencies or intensities.
Timed content cannot be paused
Moving, auto-updating, or timed embedded content cannot be paused, stopped, or controlled.
Security and Permissions
5 checksSecurity settings block assistive technology
Permissions or encryption prevent text extraction, screen reader access, copying for accessibility, or remediation.
Document requires inaccessible authentication
Password, certificate, or launch behavior prevents accessible opening or reading.
External resources are required for meaning
The PDF depends on remote content, external files, or network access to convey accessible information.
Redacted content remains exposed
Hidden or redacted content is still present in text, tags, metadata, or alternate descriptions.
Metadata exposes sensitive information
Document metadata contains private data that should not be included in reports or remediation output.
Conformance and File Integrity
7 checksPDF/UA identifier is missing
The document does not declare PDF/UA conformance metadata.
PDF/UA identifier is present but unsupported
The conformance claim is malformed, inconsistent, or not supported by the file structure.
PDF syntax is malformed
Low-level syntax errors may prevent reliable parsing, validation, or assistive technology access.
Cross-reference table is damaged
Object references cannot be resolved reliably because xref data is missing or corrupt.
Required fonts are not embedded
Fonts needed for consistent text rendering and Unicode mapping are missing.
Generated remediation would remove existing accessibility
A future remediation pass must not discard valid tags, metadata, alt text, or form labels already present.
Validation profile cannot be determined
The PDF version, declared standards, or metadata are insufficient to choose the correct validation strategy.
Page Geometry and Rendering
12 checksPage rotation conflicts with content orientation
Page rotation metadata and visible content orientation disagree, causing awkward reading, OCR, or navigation.
Crop box hides meaningful content
The visible page crop clips text, controls, figures, or annotations that remain structurally present.
Bleed or trim boxes expose confusing content
Printer marks, bleed content, or trim artifacts are exposed in ways that can confuse reading or remediation.
Pages have inconsistent orientation without reason
Unexpected portrait and landscape changes make navigation, magnification, and reflow harder.
Pages have inconsistent size without reason
Mixed page sizes create unpredictable zoom, scrolling, and output behavior.
Scanned page is skewed
A scanned page is tilted enough to reduce readability or OCR accuracy.
Scanned page has poor exposure
A scan is too dark, too light, washed out, or noisy for comfortable reading or reliable OCR.
Content appears outside page bounds
Text, annotations, or tagged objects are positioned off the visible page or partly outside readable bounds.
Layer visibility hides meaningful content
Optional content groups or layers hide information that remains necessary for understanding.
Layer order changes meaning
Stacking order causes labels, highlights, or annotations to obscure or alter the meaning of content.
Transparent content reduces readability
Opacity, blend modes, or overlays make text and graphics hard to perceive.
Print-only marks are exposed as content
Crop marks, registration marks, color bars, or production notes are included in the accessible reading model.
User Experience and Understandability
20 checksFile name is not meaningful
The PDF file name does not help users identify the document, version, language, or purpose.
Document purpose is unclear at the start
The opening content does not quickly explain what the document is for or who should use it.
Long document lacks table of contents
A long or complex document has no usable contents list for orientation and navigation.
Table of contents links are missing or broken
Contents entries do not link to the correct sections or cannot be activated reliably.
Section summaries are missing where needed
Dense or complex sections lack summaries that help users understand key points quickly.
Plain language risk is high
Sentences, terminology, or structure may be unnecessarily hard to understand for the intended audience.
Undefined jargon or abbreviations are present
Specialized terms, acronyms, or internal labels are used without nearby explanation.
Instructions are not step-based
Procedures or tasks are written as dense prose instead of clear ordered steps.
Critical deadlines are not easy to find
Dates, deadlines, or time-sensitive actions are buried in surrounding text.
Required actions are not explicit
Users cannot easily determine what they need to do, by when, and through which channel.
Contact options are missing
The document does not provide a clear way to ask questions, request help, or get an accessible alternative.
Contact details are not actionable
Email addresses, phone numbers, or URLs are plain text when they should be usable links.
Document version is unclear
Users cannot tell whether the PDF is current, draft, archived, or superseded.
Language versions are not linked
Alternative language versions exist but are not discoverable from the document.
Accessible alternative format is not offered where needed
A complex or hard-to-remediate PDF does not tell users how to get an accessible alternative.
Dense page layout increases cognitive load
Crowded content, weak grouping, or poor spacing makes the document harder to scan and understand.
Visual grouping is not conveyed structurally
Cards, panels, callouts, or grouped fields rely on visual proximity without structural relationships.
Warnings are not clearly distinguished
Important notices, warnings, or errors do not have clear text and structural treatment.
Status information is unclear
Statuses such as approved, rejected, overdue, optional, or complete are not explicit in text.
Document relies on spatial instructions
Instructions such as see left, below, red box, or above are used without structural or textual alternatives.
Magnification and Small Screen Use
10 checksContent cannot be understood when zoomed
At high magnification, relationships between labels, values, captions, and sections become hard to follow.
Horizontal scrolling is excessive
Users must pan back and forth repeatedly to read normal text at larger zoom levels.
Line length is too long
Paragraph lines are long enough to make tracking, magnification, and reading harder.
Important content depends on fixed two-page view
Meaning relies on seeing facing pages together, which may fail on small screens or assistive setups.
Footnotes are too far from references
Notes are difficult to use when zoomed because references and note text are separated without navigation support.
Callouts detach from referenced content
Magnification or small-screen viewing makes callouts, labels, or legends hard to connect with their targets.
Touch targets are too small
Interactive annotations, buttons, or form controls are difficult to activate accurately on touch devices.
Touch targets are too close together
Adjacent interactive areas increase the risk of accidental activation.
Reflow order differs from logical order
Viewer reflow presents content in a different sequence from the tagged structure.
Zoomed form completion is difficult
Labels, fields, help text, and errors cannot be kept in context while filling out the form at high zoom.
Consistency and Predictability
8 checksSimilar headings use inconsistent levels
Sections with the same visual role are exposed at different heading levels.
Similar components use inconsistent tags
Repeated callouts, cards, lists, or tables are tagged differently across the document.
Repeated links have inconsistent names
Links to the same destination use different accessible names without a meaningful reason.
Same link text points to different destinations
Identical link names lead to different locations or actions, creating ambiguity.
Form field naming is inconsistent
Fields that ask for similar information use inconsistent names, labels, or instructions.
Status labels are inconsistent
The same status or category is described with different words, icons, or colors across the document.
Page header structure changes unexpectedly
Repeated page regions appear in inconsistent positions or with inconsistent artifact treatment.
Navigation labels are inconsistent
Bookmarks, table of contents entries, headings, and page labels do not use consistent naming.
Human Review Signals
10 checksAlternative text needs human judgment
The image purpose is contextual enough that deterministic checks cannot confirm whether the description is sufficient.
Reading order needs visual confirmation
The layout is complex enough that tag order should be compared with page screenshots.
Table relationships need human confirmation
Complex, irregular, or nested table structures need review beyond simple header detection.
Chart description needs subject matter review
A chart summary may need domain knowledge to confirm that the right insight is described.
Plain language needs editorial review
The document may be understandable only after a human reviews audience, terminology, and legal or technical constraints.
Artifact decisions need confirmation
Some repeated or decorative-looking content may still be meaningful and should be reviewed before hiding it.
Form instructions need usability review
A form may be technically labeled but still confusing to complete without human usability review.
Remediation confidence is low
Automated remediation would affect meaningful structure or content and should require review before export.
Conflicting signals require review
The visual layout, OCR text, metadata, and structure tree disagree about the content meaning.
Legal or compliance wording may need preservation
Suggested changes could alter regulated, contractual, or legal text and should be reviewed by an owner.
Remediation Quality
12 checksRemediation changes visible layout unexpectedly
Fixing tags or metadata should not alter the visible document unless explicitly requested.
Remediation changes text content unexpectedly
Automated fixes must not rewrite, delete, duplicate, or reorder visible text unintentionally.
Remediation breaks existing links
Fixes must preserve link targets, destinations, and annotation actions.
Remediation breaks form behavior
Fixes must preserve form field values, validation, calculation, submission, and keyboard behavior.
Remediation removes digital signatures
Output generation must warn before invalidating or removing signatures, seals, or certification.
Remediation invalidates conformance metadata
Output metadata must match the actual accessibility and PDF standards state after remediation.
Remediation creates duplicate tags
Automated tagging should not duplicate content in the structure tree or reading stream.
Remediation loses document metadata
Title, author, subject, language, identifiers, and other important metadata must be preserved or intentionally updated.
Remediation increases file size excessively
Output should avoid unnecessary rasterization, embedded duplicates, or bloated resources that harm download and use.
Remediation output cannot be reopened reliably
The remediated PDF must remain valid and readable in common PDF viewers and validators.
Remediation report lacks evidence
Findings and fixes should include enough evidence for a user to understand what changed and why.
Remediation confidence is not reported
Automated decisions should expose confidence, uncertainty, and whether human review is recommended.
Export and Delivery
8 checksOutput file name is ambiguous
The exported file name does not distinguish remediated, reviewed, draft, or original versions clearly.
Original and remediated files can be confused
The workflow does not make it clear which file is original, remediated, verified, or ready to publish.
Accessibility report is not accessible
The generated report itself must be structured, readable, and accessible.
Report findings are not prioritized
Users cannot distinguish blocking issues, warnings, review items, and informational results.
Report lacks affected page references
Findings do not identify the relevant page, object, annotation, or structure element.
Report lacks before-and-after comparison
Users cannot inspect what changed between the original and remediated PDF.
Report omits unresolved review items
Human-review findings must remain visible until accepted, dismissed, or resolved.
Output is not checked after remediation
The exported PDF must be revalidated after fixes rather than assuming remediation succeeded.