acc-assess

WCAG 2.2 + WAI-ARIA APG

acc-assess

A set of skills that validate accessibility based on user journeys, using deterministic tools and agents. axe-core static analysis, AbleDOM real-time analysis and WAI-ARIA APG pattern review against WCAG 2.2 combine into a comprehensive analysis.

It still does not replace manual testing. It aims to make manual testing more accurate. Research phase: no evals yet, and validated with few models so far.

Report summary: half-donut charts for issues and criteria, issue groups by severity and a strip of screen thumbnails.

Install

Four Agent Skills plus a Claude Code plugin wrapper. Install all four together: the companion skills load Playwright and shared code from accessibility-assessment. Needs Node.js 20+, npm, and network access on first setup.

  • Option 1 · skills CLI recommended
    Vercel's skills CLI installs into the native skill folders of Claude Code, Cursor, Codex, GitHub Copilot, Gemini CLI, Antigravity and more. Omit --global for a project install; add --agent claude-code cursor to pick agents.
    Commands
    npx skills add jurokapsiar/acc-assess --skill '*' --global --yes
    cd ~/.agents/skills/accessibility-assessment   # or the path the CLI printed
    npm run setup
  • Option 2 · Claude Code plugin auto-updates
    Setup is automatic: a background SessionStart hook installs whatever is missing in the installed plugin copy, and again after each plugin update. Skills are namespaced, e.g. /acc-assess:accessibility-assessment.
    Commands
    /plugin marketplace add jurokapsiar/acc-assess
    /plugin install acc-assess@acc-assess
  • Option 3 · Other agents manual
    Copy all four skills/*/ folders, side by side, into the agent's documented skill directory, then run setup in accessibility-assessment/.
    Commands
    cp -r skills/* <agent-skills-dir>/
    cd <agent-skills-dir>/accessibility-assessment
    npm run setup
  • Option 4 · Release download planned
    A GitHub Action will publish a skills/ zip with every version tag (tests and plugin validation first). Until the first release, download the repository ZIP from GitHub and copy the four skills/*/ folders as in option 3.
    Commands
    cd ~/.claude/skills/accessibility-assessment
    npm run setup

The skills

Ask in plain language. Each skill knows its scripts and hands off to the next.

scope-definition

Turns a sentence into an approved, reproducible test scope.

  • Screens, states and viewports from your prompt
  • User journeys as replayable steps
  • Pause for human approval before any assessment

Assess the checkout flow on mobile and desktop.

Output: scope.json, journeys.json

accessibility-assessment

Runs the tools, adds agent review, and writes the report and fix files.

  • axe-core, AbleDOM and Playwright checks per state
  • Agent judgement for what tools cannot decide
  • One fix file per issue, with repro and verify command

Run an accessibility assessment of the approved scope.

Output: report.html, fix-*.md

fix-verify

Proves that an applied fix worked, without new regressions.

  • Re-runs only what the fix file names
  • Before/after crops for every occurrence
  • Status: fixed, not fixed, regressed, unverified, needs review

Verify the fixes I just applied.

Output: fix-report.html

screen-reader-validation

Listens to real NVDA or VoiceOver output for each journey.

  • Guidepup drives the reader with OS-level keys
  • Session log: step, keys, what was heard
  • Silent, missing or wrong announcements become issues

Check the new-task journey with NVDA.

Output: sr-report.html

Workflow

Seven steps, four skills. Each step writes a file the next step reads, so you can stop, inspect and resume.

Workflow diagramSeven steps from left to right: 1 Scope, 2 Inventory, 3 Assess, 4 Report, 5 Fix files, 6 Fix-verify, 7 Screen reader. A human approves the scope. Code edits after step 5 are out of scope due to code base specifics. When fix-verify reports not fixed, the loop goes back to step 5.scope-definitionaccessibility-assessmentfix-verifyscreen-reader-validation 1 Scope scope.json · journeys 2 Inventory inventory.json 3 Assess axe · AbleDOM · Playwright 4 Report report.html 5 Fix files fix-*.md 6 Fix-verify fix-report.html 7 Screen reader NVDA · VoiceOver not fixed → revise human approves Code edit out of scopeWorkflow diagramSeven steps from top to bottom: 1 Scope, 2 Inventory, 3 Assess, 4 Report, 5 Fix files, 6 Fix-verify, 7 Screen reader. When fix-verify reports not fixed, the loop goes back to step 5. 1 Scope scope.json · journeysscope-def. 2 Inventory inventory.jsonassessment 3 Assess axe · AbleDOM · Playwrightassessment 4 Report report.htmlassessment 5 Fix files fix-*.mdassessment 6 Fix-verify fix-report.htmlfix-verify 7 Screen reader NVDA · VoiceOversr-valid. notfixed
  1. Define the scope

    you approve

    Describe what to test in plain words. The skill turns it into screens, states and user journeys and asks you to approve them.

    node scripts/scope.mjs init --prompt "New task flow and settings" --url http://localhost:3000
    Skill: scope-definitionWrites: scope.json journeys.json
  2. Discover the inventory

    skill

    Features, screens and states are discovered and checked against the scope, so nothing in scope is skipped silently.

    node scripts/inventory.mjs validate
    node scripts/inventory.mjs verify
    Skill: accessibility-assessmentWrites: inventory.json
  3. Assess

    skill

    Every screen and state at each viewport: axe-core, AbleDOM, and Playwright checks such as keyboard walk, overlays, 320px reflow and text spacing.

    node scripts/assess.mjs --scope .a11y/scope.json
    Skill: accessibility-assessmentWrites: results.json screenshots
  4. Review and report

    skill + agent

    The agent reviews what tools cannot decide and records its judgement. The report groups failures, items to review and suggestions.

    node scripts/report.mjs
    Skill: accessibility-assessmentWrites: report.html
  5. Write fix files

    code edit out of scope

    One Markdown file per issue: the fix, the changes, the occurrences, repro steps and a verify command. Applying the change is out of scope due to code base specifics.

    node scripts/fixes.mjs
    Skill: accessibility-assessmentWrites: fix-*.md
  6. Verify the fix

    skill

    Re-runs each fix file against the changed app and proves it with before/after crops of every occurrence.

    node scripts/verify.mjs .a11y/fixes/fix-*.md
    Skill: fix-verifyWrites: fix-report.html exit 0 · 1 · 2 · 3
  7. Validate with a screen reader

    skill

    Drives NVDA (Windows) or VoiceOver (macOS) through the same journeys with Guidepup and checks what is actually announced.

    node scripts/sr.mjs preflight
    node scripts/sr.mjs run --scope .a11y/scope.json
    node scripts/sr.mjs analyse && node scripts/sr.mjs report
    Skill: screen-reader-validationWrites: sr-report.html

The skills never edit your source code. They read, run and report. Fixes are proposals in fix-*.md. Editing code is out of scope due to code base specifics; apply them with your own process, then fix-verify proves the result.

Interaction modalities

Every journey step records how it was performed, so a run can't pass just because the runner clicked through. Described as specified; names and flags may still be refined.

keyboard

Tab, Shift+Tab, arrows, Enter, Space. Other keys only by escalation.

Phase
assessment
Rules
kbd-*
Judged
Mouse or programmatic focus needed = failure.

mouse

Pointer click, hover and drag. Keys only to type values.

Phase
assessment
Rules
ptr-*
Judged
The reference pass: proves the journey is reachable at all.

sr-keyboard

The same keys, sent through the reader as OS keys. Browse or focus mode is recorded.

Phase
screen reader
Rules
kbd-* + sr-*
Judged
Same failures, with reader context.

sr-navigation

Reader commands: quick-nav keys, elements list or rotor, browse arrows; keyboard keys too.

Phase
screen reader
Rules
srnav-*
Judged
Reachable only by keyboard = review.

Run with --modality keyboard,mouse (assessment) or --modality sr-keyboard,sr-navigation (screen reader); all runs both of a skill's phases. Each modality is a separate replay in a fresh browser. Journeys without modalities keep the default behaviour.

When the keyboard gets stuck: the escalation ladder

The runner climbs only as far as needed, records each rung, and keeps going.

  1. rung 0Core keysTab Shift+Tab ↑ ↓ ← → Enter Space
    no issue
  2. rung 1Other keysEsc Home End F6 or a documented shortcut
    kbd-nonstandard-key · suggestion
  3. rung 2Programmatic focusThe runner had to call focus()
    kbd-programmatic-focus · failure
  4. rung 3MouseOnly a click got through
    kbd-mouse-required · failure
  5. rung 4Give upThe journey is reported as partial
    partial journey

Why it got stuck

  • trap
  • not-in-tab-sequence
  • composite-no-arrows
  • no-activation
  • focus-lost
  • target-hidden

Each reason maps to a kbd-* rule with evidence and a fix.

Screen-reader navigation order

  1. Quick-nav: H D K B F …
  2. Elements list: NVDA+F7 or the VoiceOver rotor
  3. Browse-mode arrows and reading
  4. Keyboard keys (then it's srnav-* review)

Recorded per step

Intent

  • reach
  • activate
  • enter-text
  • choose
  • toggle
  • dismiss
  • open
  • read

Focus came from

  • key
  • pointer
  • page
  • programmatic

18 modality rules are in the catalog below. Page autofocus isn't a runner violation; when it leaves the reader in focus mode it is flagged for review.

The reports

Static HTML that opens from disk. Every finding links to a screen, a state, a viewport, a crop and a source line.

Assessment report summary section.

Summary

  1. Issues by severityFailures as a half-donut, split by impact.
  2. CriteriaCriteria with issues, needing review, or with nothing detected. Never "pass": tools can only fail to find problems.
  3. Screens with issuesHow many scoped screens have at least one failure.
  4. IssuesConfirmed failures, grouped by rule.
  5. Needs reviewTool results that need judgement, and agent findings.
  6. ScreensThumbnails jump to every state assessed.
One issue card from the report.

Issue card

  1. ChipsKind, impact, WCAG criterion and the rule that found it.
  2. FixWhat to change, in plain words, with the affected files.
  3. SourceFile and line resolved from the rendered element.
  4. OccurrenceScreen, state and viewport. Steps to reproduce are one click away.
  5. Element cropThe failing element, outlined on the real screenshot.

Report sections

Each section has its own tint, used in the title row of every box.

Fix-verify report

Fixed, not fixed, regressed, unverified or needs review, for each fix file.

Verification status

Fix-verify summary: one of one fix verified as fixed; eight of eight proof elements clear.

Exit code 0 when all fixed; 1, 2 or 3 otherwise.

Proof

Fix proof: before and after crops of the same element, now reported clear.

Before and after, element by element.

Screen-reader report

Real NVDA or VoiceOver output, step by step.

Summary

Screen-reader report summary: issues by severity and how many steps were narrated.

Issues and how many steps produced speech.

Finding

Screen-reader finding: navigation produced no announcement.

What was heard, what was expected, and the fix.

Example issues

One from each assessor. Every issue carries the evidence, the steps to reproduce it and a fix you can apply.

Accessible name does not contain the visible label

  • axe-core
  • failure
  • serious
  • SC 2.5.3
  • label-content-name-mismatch

A date button shows “Tuesday, November 3” but its accessible name is “Change task date”. Speech-input users who say what they see can’t activate it.

Steps to reproduce

  1. Open the app at 1280×800.
  2. Select New task.
  3. Inspect the date button in the dialog.
  4. Compare its visible text with its accessible name.
Fix

Start the accessible name with the visible text.

- <button aria-label="Change task date">Tuesday, November 3</button>+ <button aria-label="Tuesday, November 3, change date">Tuesday, November 3</button>
Task dialog with two numbered red boxes around the date and time buttons.
Fix-verify proof for this issue: before and after crops, now clear.

Toggle field name needs a check

  • axe-core
  • review
  • serious
  • SC 4.1.2
  • aria-toggle-field-name

axe-core can’t tell if a wrapping label names a role="switch". The agent checks each occurrence and records the result; mismatches become their own finding.

Steps to reproduce

  1. Open the switch demo at 1280×800.
  2. Inspect each switch.
  3. Compare the label text with the computed name.
Fix

Keep aria-label equal to the visible label, or reference it with aria-labelledby.

+ <span role="switch" aria-labelledby="lbl-apca" aria-checked="true">
Report card for the toggle field review item with fix text, source and element crops.

Tab order does not follow reading order

  • agent
  • review
  • SC 1.3.2, 2.4.3
  • tab-order-reading-order

The keyboard walk is overlaid on the screenshot as numbered stops. The agent compares that order with the visual layout.

Steps to reproduce

  1. Open the screen at 1280×800.
  2. Press Tab from the top of the page.
  3. Follow the numbered stops.
Fix

Make DOM order match visual order. Avoid positive tabindex and CSS-only reordering.

Screenshot with numbered tab stops drawn over each focusable element.

Navigation produced no announcement

  • Guidepup
  • review
  • moderate
  • SC 2.4.3
  • sr-silent-navigation

After a route change the screen reader said nothing, so the user can’t tell where they are.

Heard
(nothing)
Expected
The new page title or heading

Steps to reproduce

  1. Start NVDA.
  2. Run the journey with sr.mjs run.
  3. Activate the link that changes the view.
  4. Listen.
Fix

Give every page/view a unique <title> and move focus to the new view's heading (or announce the route change).

Screen-reader report finding with heard and expected announcements.

Step can only be completed with a mouse

  • Playwright
  • failure
  • critical
  • SC 2.1.1
  • kbd-mouse-required
  • modality keyboard

In the keyboard phase the runner climbed the escalation ladder; only a pointer click completed the step.

Steps to reproduce

  1. Run the journey with --modality keyboard.
  2. Tab to the control.
  3. Press Enter or Space.
Fix

Make the control operable from the keyboard: use a native button/link/input, or add tabindex="0" with Enter/Space handlers and a matching role.

- <div class="add" onClick={add}>Add item</div>+ <button type="button" onClick={add}>Add item</button>

Escalation trace

step 3  intent: activate  target: "Add item"
  rung 0  Tab ×14, Enter, Space   → not reached
  rung 1  Escape, Home, End        → not reached
  rung 2  programmatic focus()     → focused, Enter ignored
  rung 3  mouse click              → completed
  stuck: no-activation   focus came from: programmatic

Criteria catalog

Every WCAG 2.2 criterion, rule and APG pattern the suite covers, and what assesses it. Generated from the rule sources.

Rules by assessor

182 rules
  • axe-core103
  • AbleDOM7
  • Playwright27
  • Guidepup21
  • agent24

Criteria by level

86 criteria
  • Level A31
  • Level AA24
  • Level AAA31

How criteria are covered

86 criteria
  • Automated + agent41
  • Agent only31
  • Manual only14

Criteria per assessor

  • axe-core31 criteria
  • AbleDOM7 criteria
  • Playwright8 criteria
  • Guidepup8 criteria
  • agent55 criteria
  • manual21 criteria

A criterion can have several assessors.

  • axe-coreAutomated DOM rules per state and viewport.
  • AbleDOMFocus and accessibility-tree checks at runtime.
  • PlaywrightCustom checks: keyboard walk, overlays, reflow, text spacing, keyboard modality.
  • GuidepupDrives NVDA or VoiceOver and checks what is announced.
  • agentLLM judgement on screenshots, DOM and tool output.
  • manualNeeds a human; not automatable.

Criteria
SCCriterionLevelPrincipleAssessed byRules
1.1.1Non-text ContentAPerceivable
  • axe-core
  • Guidepup
  • agent
8 rules
  • aria-meter-name
  • aria-progressbar-name
  • image-alt
  • input-image-alt
  • object-alt
  • role-img-alt
  • svg-img-alt
  • sr-symbol-noise
1.2.1Audio-only and Video-only (Prerecorded)APerceivable
  • axe-core
  • manual
1 rule
  • audio-caption
1.2.2Captions (Prerecorded)APerceivable
  • axe-core
  • manual
1 rule
  • video-caption
1.2.3Audio Description or Media Alternative (Prerecorded)APerceivable
  • axe-core
  • manual
—
1.2.4Captions (Live)AAPerceivable
  • axe-core
  • manual
—
1.2.5Audio Description (Prerecorded)AAPerceivable
  • axe-core
  • manual
—
1.2.6Sign Language (Prerecorded)AAAPerceivable
  • manual
—
1.2.7Extended Audio Description (Prerecorded)AAAPerceivable
  • manual
—
1.2.8Media Alternative (Prerecorded)AAAPerceivable
  • manual
—
1.2.9Audio-only (Live)AAAPerceivable
  • manual
—
1.3.1Info and RelationshipsAPerceivable
  • axe-core
  • AbleDOM
  • Guidepup
  • agent
14 rules
  • aria-hidden-body
  • aria-required-children
  • aria-required-parent
  • definition-list
  • dlitem
  • list
  • listitem
  • p-as-heading
  • table-fake-caption
  • td-has-header
  • td-headers-attr
  • th-has-data-cells
  • missing-id
  • sr-no-headings
1.3.2Meaningful SequenceAPerceivable
  • Guidepup
  • agent
2 rules
  • sr-reading-order
  • tab-order-reading-order
1.3.3Sensory CharacteristicsAPerceivable
  • agent
—
1.3.4OrientationAAPerceivable
  • axe-core
  • agent
2 rules
  • css-orientation-lock
  • orientation-lock
1.3.5Identify Input PurposeAAPerceivable
  • axe-core
  • agent
1 rule
  • autocomplete-valid
1.3.6Identify PurposeAAAPerceivable
  • agent
—
1.4.1Use of ColorAPerceivable
  • axe-core
  • agent
2 rules
  • link-in-text-block
  • color-only-state
1.4.2Audio ControlAPerceivable
  • axe-core
1 rule
  • no-autoplay-audio
1.4.3Contrast (Minimum)AAPerceivable
  • axe-core
  • AbleDOM
  • agent
3 rules
  • color-contrast
  • contrast
  • text-on-image-contrast
1.4.4Resize TextAAPerceivable
  • axe-core
  • agent
  • manual
2 rules
  • meta-viewport
  • text-resize-clipping
1.4.5Images of TextAAPerceivable
  • agent
1 rule
  • image-of-text
1.4.6Contrast (Enhanced)AAAPerceivable
  • axe-core
1 rule
  • color-contrast-enhanced
1.4.7Low or No Background AudioAAAPerceivable
  • manual
—
1.4.8Visual PresentationAAAPerceivable
  • agent
—
1.4.9Images of Text (No Exception)AAAPerceivable
  • agent
—
1.4.10ReflowAAPerceivable
  • Playwright
  • agent
1 rule
  • reflow-horizontal-scroll
1.4.11Non-text ContrastAAPerceivable
  • agent
1 rule
  • non-text-contrast
1.4.12Text SpacingAAPerceivable
  • axe-core
  • Playwright
  • agent
2 rules
  • avoid-inline-spacing
  • text-spacing-clipping
1.4.13Content on Hover or FocusAAPerceivable
  • agent
1 rule
  • tooltip-dismiss-hover
2.1.1KeyboardAOperable
  • axe-core
  • AbleDOM
  • Playwright
  • agent
11 rules
  • frame-focusable-content
  • scrollable-region-focusable
  • server-side-image-map
  • focus-lost
  • not-keyboard-focusable
  • kbd-mouse-required
  • kbd-programmatic-focus
  • kbd-not-focusable
  • kbd-no-activation
  • kbd-composite-no-arrows
  • apg-keyboard-support
2.1.2No Keyboard TrapAOperable
  • Playwright
2 rules
  • keyboard-trap
  • kbd-trap
2.1.3Keyboard (No Exception)AAAOperable
  • axe-core
1 rule
  • scrollable-region-focusable
2.1.4Character Key ShortcutsAOperable
  • agent
1 rule
  • character-key-shortcut
2.2.1Timing AdjustableAOperable
  • axe-core
  • manual
1 rule
  • meta-refresh
2.2.2Pause, Stop, HideAOperable
  • axe-core
  • agent
2 rules
  • blink
  • marquee
2.2.3No TimingAAAOperable
  • manual
—
2.2.4InterruptionsAAAOperable
  • axe-core
1 rule
  • meta-refresh-no-exceptions
2.2.5Re-authenticatingAAAOperable
  • manual
—
2.2.6TimeoutsAAAOperable
  • manual
—
2.3.1Three Flashes or Below ThresholdAOperable
  • manual
—
2.3.2Three FlashesAAAOperable
  • manual
—
2.3.3Animation from InteractionsAAAOperable
  • manual
—
2.4.1Bypass BlocksAOperable
  • axe-core
  • Guidepup
  • agent
2 rules
  • bypass
  • sr-no-landmarks
2.4.2Page TitledAOperable
  • axe-core
  • agent
2 rules
  • document-title
  • page-title-not-distinct
2.4.3Focus OrderAOperable
  • AbleDOM
  • Playwright
  • Guidepup
  • agent
12 rules
  • focus-lost
  • bad-focus
  • focus-not-moved-into-dialog
  • initial-focus-outside-overlay
  • focus-escapes-modal
  • focus-not-restored
  • sr-focus-not-in-dialog
  • sr-focus-not-returned
  • sr-silent-navigation
  • kbd-programmatic-focus
  • kbd-focus-lost
  • tab-order-reading-order
2.4.4Link Purpose (In Context)AOperable
  • axe-core
  • agent
2 rules
  • area-alt
  • link-name
2.4.5Multiple WaysAAOperable
  • agent
1 rule
  • single-navigation-path
2.4.6Headings and LabelsAAOperable
  • AbleDOM
  • agent
2 rules
  • focusable-element-label
  • ambiguous-heading-label
2.4.7Focus VisibleAAOperable
  • AbleDOM
  • Playwright
  • agent
3 rules
  • bad-focus
  • focus-not-visible
  • focus-indicator-per-theme
2.4.8LocationAAAOperable
  • agent
—
2.4.9Link Purpose (Link Only)AAAOperable
  • axe-core
1 rule
  • identical-links-same-purpose
2.4.10Section HeadingsAAAOperable
  • agent
—
2.4.11Focus Not Obscured (Minimum)AAOperable
  • Playwright
1 rule
  • focus-obscured
2.4.12Focus Not Obscured (Enhanced)AAAOperable
  • Playwright
1 rule
  • focus-partially-obscured
2.4.13Focus AppearanceAAAOperable
  • agent
—
2.5.1Pointer GesturesAOperable
  • agent
1 rule
  • drag-pointer-alternative
2.5.2Pointer CancellationAOperable
  • manual
—
2.5.3Label in NameAOperable
  • axe-core
  • agent
1 rule
  • label-content-name-mismatch
2.5.4Motion ActuationAOperable
  • manual
—
2.5.5Target Size (Enhanced)AAAOperable
  • agent
—
2.5.6Concurrent Input MechanismsAAAOperable
  • manual
—
2.5.7Dragging MovementsAAOperable
  • agent
1 rule
  • drag-pointer-alternative
2.5.8Target Size (Minimum)AAOperable
  • axe-core
1 rule
  • target-size
3.1.1Language of PageAUnderstandable
  • axe-core
3 rules
  • html-has-lang
  • html-lang-valid
  • html-xml-lang-mismatch
3.1.2Language of PartsAAUnderstandable
  • axe-core
  • agent
1 rule
  • valid-lang
3.1.3Unusual WordsAAAUnderstandable
  • agent
—
3.1.4AbbreviationsAAAUnderstandable
  • agent
—
3.1.5Reading LevelAAAUnderstandable
  • agent
—
3.1.6PronunciationAAAUnderstandable
  • agent
—
3.2.1On FocusAUnderstandable
  • agent
1 rule
  • unexpected-context-change
3.2.2On InputAUnderstandable
  • agent
1 rule
  • unexpected-context-change
3.2.3Consistent NavigationAAUnderstandable
  • agent
1 rule
  • inconsistent-navigation-order
3.2.4Consistent IdentificationAAUnderstandable
  • agent
1 rule
  • inconsistent-control-identification
3.2.5Change on RequestAAAUnderstandable
  • axe-core
1 rule
  • meta-refresh-no-exceptions
3.2.6Consistent HelpAUnderstandable
  • agent
1 rule
  • inconsistent-help-location
3.3.1Error IdentificationAUnderstandable
  • Guidepup
  • agent
2 rules
  • sr-error-not-announced
  • error-identification-suggestion
3.3.2Labels or InstructionsAUnderstandable
  • axe-core
  • agent
1 rule
  • form-field-multiple-labels
3.3.3Error SuggestionAAUnderstandable
  • agent
1 rule
  • error-identification-suggestion
3.3.4Error Prevention (Legal, Financial, Data)AAUnderstandable
  • agent
—
3.3.5HelpAAAUnderstandable
  • agent
—
3.3.6Error Prevention (All)AAAUnderstandable
  • agent
—
3.3.7Redundant EntryAUnderstandable
  • agent
1 rule
  • repeated-input-request
3.3.8Accessible Authentication (Minimum)AAUnderstandable
  • agent
—
3.3.9Accessible Authentication (Enhanced)AAAUnderstandable
  • agent
—
4.1.2Name, Role, ValueARobust
  • axe-core
  • AbleDOM
  • Guidepup
  • agent
39 rules
  • area-alt
  • aria-allowed-attr
  • aria-braille-equivalent
  • aria-command-name
  • aria-conditional-attr
  • aria-deprecated-role
  • aria-hidden-body
  • aria-hidden-focus
  • aria-input-field-name
  • aria-prohibited-attr
  • aria-required-attr
  • aria-roledescription
  • aria-roles
  • aria-tab-name
  • aria-toggle-field-name
  • aria-tooltip-name
  • aria-valid-attr
  • aria-valid-attr-value
  • button-name
  • duplicate-id-aria
  • frame-title
  • frame-title-unique
  • input-button-name
  • input-image-alt
  • label
  • link-name
  • nested-interactive
  • select-name
  • summary-name
  • focusable-element-label
  • focusable-in-atomic
  • missing-id
  • sr-blank-announcement
  • sr-control-without-name
  • sr-missing-role
  • sr-missing-state
  • sr-dialog-not-announced
  • sr-silent-action
  • apg-role-state
4.1.3Status MessagesAARobust
  • Guidepup
  • agent
3 rules
  • sr-status-not-announced
  • sr-error-not-announced
  • status-not-announced