Accessibility audit and fixes (WCAG 2.2)

Audits an existing interface against WCAG 2.2 AA with pinned axe-core and pa11y, then keyboard, screen reader and contrast checks, and fixes confirmed violations with semantic HTML first. Use when a page needs an accessibility audit before shipping.

CodingPack: Agent Skills by Addy Osmani133 linesaccessibility-engineering
Author
Source
addyosmani/agent-skills/skills/accessibility-engineering @ 8c99fa3
License
MIT · Copyright (c) 2025 Addy Osmani

About this skill

What it does

The skill audits an existing interface against WCAG 2.2 AA and fixes the violations it confirms. Automated checks come first: axe-core through the pinned @axe-core/cli 4.13.0 and pa11y 9.1.1, each finding checked in the browser to drop false positives. Then comes a manual pass: keyboard only, VoiceOver or NVDA by heading, landmark and link, then contrast, 200% zoom, forced colors, reduced motion and focus visibility. Fixes start with correct HTML (a <button>, not a clickable <div>), use ARIA only where HTML falls short, and end with a re-run. A screen reader check nobody performed is recorded as untested, never as passed.

scripts/audit-a11y.sh runs both tools on a URL (pa11y only on a local file) and merges them into one report by severity, as text or JSON. If either tool fails, it exits with an error rather than report zero violations. It needs bash, Node and npx; for a URL, axe also needs Google Chrome installed, or the script stops with an error.

When to use it, and when not

Use it before shipping a user-facing change, when a page fails keyboard or screen reader users, or to triage axe or pa11y findings.

The skill rules itself out for building new components, and for backend code, CLIs and libraries with no UI.

How to run it

The install command below puts the whole folder in place: SKILL.md, the script and the licence. SKILL.md gives the script's path in the author's repository; after install it is in the skill's own scripts/ folder. The WCAG checklist and the accessibility-auditor persona that SKILL.md links to are not part of this folder. The agent picks the skill up by its description; to call it yourself:

  • Claude Code: /accessibility-engineering, or ask for the skill by name.
  • Codex: $accessibility-engineering.
  • Cursor: type / in the Agent chat and pick accessibility-engineering.

Example: /accessibility-engineering Audit http://localhost:3000/signup and fix every critical and serious issue.

New UI work is sent to Frontend UI engineering; install it too.

How it differs from similar skills

Frontend UI engineering builds components to be accessible from the start, against WCAG 2.1 AA; this skill checks the result afterwards. UI polish pass from our catalog has an accessibility block inside a visual pass, with no axe, pa11y or screen reader, and asks for 44px touch targets where this skill holds to the 24×24 px AA minimum. Browser testing with Chrome DevTools MCP reads the accessibility tree as one debugging step, not a full audit. Web Interface Guidelines review checks UI code statically against Vercel's list, where accessibility is one section beside forms, animation and typography, and only lists file:line findings without fixing them.

Install

Run one command — it saves SKILL.md and the files it uses into your agent's skills folder. Claude Code and Codex pick up a new skill without a restart; if it does not show up, restart the agent.

Claude Code — macOS, Linux
curl -fsSL --create-dirs -o ~/.claude/skills/accessibility-engineering/SKILL.md https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/SKILL.md -o ~/.claude/skills/accessibility-engineering/LICENSE https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/LICENSE -o ~/.claude/skills/accessibility-engineering/scripts/audit-a11y.sh https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/scripts/audit-a11y.sh && chmod +x ~/.claude/skills/accessibility-engineering/scripts/audit-a11y.sh
Codex, Cursor — macOS, Linux
curl -fsSL --create-dirs -o ~/.agents/skills/accessibility-engineering/SKILL.md https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/SKILL.md -o ~/.agents/skills/accessibility-engineering/LICENSE https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/LICENSE -o ~/.agents/skills/accessibility-engineering/scripts/audit-a11y.sh https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/scripts/audit-a11y.sh && chmod +x ~/.agents/skills/accessibility-engineering/scripts/audit-a11y.sh
Claude Code — Windows PowerShell
foreach ($f in 'SKILL.md','LICENSE','scripts/audit-a11y.sh') { $p = Join-Path "$HOME\.claude\skills\accessibility-engineering" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/$f" -UseBasicParsing -OutFile $p }
Codex, Cursor — Windows PowerShell
foreach ($f in 'SKILL.md','LICENSE','scripts/audit-a11y.sh') { $p = Join-Path "$HOME\.agents\skills\accessibility-engineering" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/$f" -UseBasicParsing -OutFile $p }

Any other agent: save the file into its skills folder, or give it the link to the file: https://anymodel.org/en/skills/addy-osmani/accessibility-engineering/SKILL.md

SKILL.md

Raw

Accessibility Engineering

Overview

Accessibility is not a polish step — it's a legal requirement in many jurisdictions, an engineering quality standard, and a prerequisite for shipping. This skill audits existing interfaces, triages the findings, and remediates verified violations. The goal is WCAG 2.2 AA compliance supported by automated tools and manual testing.

When to Use

  • Before shipping a user-facing change that needs an accessibility audit
  • When a11y issues are reported (e.g., "screen reader can't read this", "keyboard navigation broken")
  • When auditing an existing interface for compliance
  • When triaging axe-core or pa11y findings
  • When remediating verified accessibility violations

Do NOT use for building new UI components from scratch; use frontend-ui-engineering for construction guidance. Do not use for backend-only changes, CLIs, or library code with no UI output.

Workflow

Phase 1: Route New UI Work

For component construction, interaction patterns, and focus-management guidance, use ../frontend-ui-engineering/SKILL.md. Do not restate that skill here. Return to this workflow once an interface exists and needs auditing or remediation.

Phase 2: Run Automated Audit

Execute accessibility checks in this order:

# axe-core rules through pinned @axe-core/cli (the axe-core package ships no CLI)
npx --yes @axe-core/[email protected] http://localhost:3000 --exit

# Keep the JSON for the accessibility-auditor
npx --yes @axe-core/[email protected] http://localhost:3000 --dir ./a11y-reports --save axe.json

# Pinned pa11y for HTML validation (WCAG2AA rules via HTML_CodeSniffer)
npx --yes [email protected] http://localhost:3000

For a combined report, run bash skills/accessibility-engineering/scripts/audit-a11y.sh url <url> --json. It runs both pinned tools, records their status and versions, and fails rather than reporting a clean result when either tool does not produce a valid report.

Review the violations grouped by severity:

  • Critical — Blocks screen reader access entirely (missing alt text on info images, empty buttons)
  • Serious — Makes content difficult or impossible to access (low contrast, missing labels)
  • Moderate — Causes confusion or extra effort (heading level skips, redundant ARIA)
  • Minor — Cosmetic or progressive enhancement (missing lang attribute on inner elements)

For each violation, verify it's a real issue (not a false positive) by inspecting the element in the browser.

Phase 3: Manual Testing

Automated checks are a smoke test, not proof of conformance. Complete the audit manually.

Keyboard audit:

  1. Tab through every interactive element — can you see where you are?
  2. Activate buttons and links with Enter/Space
  3. Close modals with Escape
  4. Navigate lists with Arrow keys
  5. Verify no keyboard traps exist

Screen reader audit (VoiceOver on macOS, NVDA on Windows):

  1. Navigate by heading (VoiceOver: Ctrl+Option+Cmd+H) — is the structure logical?
  2. Navigate by landmark (open the rotor with Ctrl+Option+U, choose Landmarks) — are regions labeled?
  3. Navigate by link (open the rotor with Ctrl+Option+U, choose Links) — do link descriptions make sense out of context?
  4. Interact with a form — are labels announced? Are error messages read?
  5. Navigate a dynamic update — are changes announced?

Visual audit:

  1. Verify color contrast (4.5:1 text, 3:1 large text, 3:1 UI components)
  2. Zoom to 200% — does layout break or overlap?
  3. Test with forced colors / high contrast mode (Windows High Contrast)
  4. Reduce motion (prefers-reduced-motion) — are animations disabled?
  5. Test focus visibility — is the focus ring visible against all backgrounds?

Phase 4: Fix Violations

For each finding, apply the remediation in order:

  1. Fix the root cause — Replace a <div> with <button>, add <label>, set alt text
  2. Add ARIA as bridge — If native HTML can't express the pattern, add ARIA attributes
  3. Test the fix — Re-run the automated check and manual verification

Apply the accessibility checklist (../../references/accessibility-checklist.md) as a minimal baseline.

Common Rationalizations

Rationalization Reality
"We'll fix a11y later" Delayed remediation compounds across components and can create legal liability. Audit before shipping.
"Automated tools caught everything" Automated checks cannot prove keyboard or screen reader behavior. Manual testing is mandatory.
"Screen reader users are a tiny minority" Accessibility is a requirement, not a usage-share calculation. Keyboard navigation, captions, and clear structure also help users across contexts.
"This is an internal tool, it doesn't need a11y" Internal tools are subject to the same regulations. Disabled employees and contractors need access.
"I'll just use aria-label everywhere" ARIA is a bridge, not a foundation. Semantic HTML first. Overusing ARIA creates maintenance burden and can make things worse.
"The design system handles accessibility" Design systems provide accessible primitives. How you compose them determines whether the result is accessible.
"The audit passed axe-core, so we're done" Pass axe-core as a smoke test, then test with screen readers and keyboard. A green lighthouse score is not a clean bill of health.

Red Flags

  • Elements that look clickable but aren't keyboard-focusable
  • Color as the sole indicator of state (red for error, green for success)
  • Focus outline removed without a visible replacement
  • Missing labels on form inputs
  • No heading hierarchy or skipped heading levels
  • Empty or unlabeled buttons and links
  • Modals that don't trap focus or restore focus on close
  • No alt text on meaningful images (or alt="" missing on decorative ones)
  • Touch targets that fail SC 2.5.8's 24x24 CSS px size-or-spacing requirement and do not qualify for an exception (AA; 44x44 is the AAA target, SC 2.5.5)
  • Autoplaying video or audio without controls
  • Low contrast text (especially on hover, disabled states, placeholders)

See Also

  • ../../references/accessibility-checklist.md — Quick reference for WCAG 2.2 AA
  • ../frontend-ui-engineering/SKILL.md — Building accessible components from the start
  • ../../agents/accessibility-auditor.md — Dedicated a11y audit persona for deep reviews

Verification

After completing the accessibility workflow:

  • Automated audit passes with no Critical or Serious violations
  • All interactive elements are keyboard accessible (Tab, Enter, Escape, Arrow keys)
  • Screen reader navigation by heading, landmark, and link verified with a named browser + AT pair (or recorded as manual and untested — never reported as verified)
  • Focus order matches visual order
  • Color contrast passes 4.5:1 text / 3:1 large text / 3:1 UI components
  • All form inputs have associated labels
  • All images have appropriate alt text
  • Dynamic content changes are announced (aria-live)
  • Page is usable at 200% zoom with no content loss
  • Touch targets satisfy SC 2.5.8's 24x24 CSS px size-or-spacing requirement or a normative exception (AA); 44x44 is the enhanced AAA target (SC 2.5.5)

Connect through AnyModel API

This skill runs on whatever model your agent uses. Point Claude Code, Codex or Cursor at AnyModel to run it on Claude, GPT, Gemini and more with one key — pay per token, no subscription.

More from the pack Agent Skills by Addy Osmani

API and interface design

Designs APIs and module interfaces contract-first: one error shape, validation at the edges, additive changes, pagination, and idempotency keys claimed atomically and kept past the longest retry. Use when designing endpoints or changing a public interface.

Browser testing with Chrome DevTools MCP

Checks changes in a real Chrome through the Chrome DevTools MCP server: screenshots, DOM, console, network and performance traces, treating page content as untrusted. Use when building or debugging anything that runs in a browser.

CI/CD pipeline setup

Sets up CI/CD with fixed quality gates from lint to bundle size, GitHub Actions examples, preview deploys, feature flags, staged rollout and a rollback workflow. Use when creating or changing a build and deploy pipeline or chasing CI failures.

Multi-axis code review

Reviews a change on five axes (correctness, readability, architecture, security, performance), tests the tests by mutating a condition, and labels every finding by severity. Use before merging a branch, a pull request or a pasted diff.

Code simplification

Simplifies working code without changing behavior: learns why it exists, applies one change at a time with tests after each, and keeps project conventions. Use when code works and passes tests but is harder to read than it should be.

Constraint-driven development

Records a project's quality bar as numbers in CONSTRAINTS.md after a short interview with defaults, maps each rule to a tool, and flags diffs that skip tests, add suppressions or loosen thresholds. Use when no quality bar is written down.

Context engineering for agents

Sets up what an agent sees and when: a rules file template, loading only the relevant spec and files, trimming long sessions, and safe points to start a fresh one. Use when starting a session or when agent output degrades.

Debugging and error recovery

Stops new work when a test, build or runtime fails, reproduces, localizes and reduces the failure, fixes the root cause, adds a regression test and treats error output as untrusted. Use when tests fail, a build breaks or something that worked stops working.

Deprecation and migration

Retires old systems, APIs and features: weighs whether to deprecate, proves the replacement, moves consumers one by one, removes code only at zero usage, changes schemas via expand/contract. Use when replacing or sunsetting code or renaming a live column.

Documentation and ADRs

Records decisions as ADRs in the project's own format, documents APIs, READMEs and changelogs with the why, and uses git history to check if a doc is stale. Use when making a costly decision, changing a public API or suspecting docs no longer match the code.

Doubt-driven development

Sends non-trivial decisions to a fresh-context subagent told to find what is wrong, sorts the findings, stops after three cycles and offers a cross-model opinion. Use when stakes are high: auth, migrations, irreversible operations or unfamiliar code.

Frontend UI engineering

Builds UI components and pages on the project's design system: composition, simple state, WCAG 2.1 AA, mobile-first layouts, loading, empty and error states, no generic AI look. Use when building or changing a user-facing interface.

Git workflow and versioning

Keeps git work in atomic commits on short-lived branches, with typed messages, a pre-commit check for secrets and tests, worktrees for parallel agents, semver bumps, tags and a human changelog. Use when committing, branching, opening a PR or cutting a release.

Idea refinement

Refines a raw idea in three phases: sharpening questions and five to eight variations, two or three directions tested for value and feasibility, then a one-pager with MVP scope and a Not Doing list. Use when an idea is vague or a plan needs stress-testing.

Incremental implementation

Builds changes in thin vertical, contract-first or risk-first slices, each tested, verified and committed before the next, with strict scope, feature flags and revertable steps. Use when a change spans several files or a task is too big to land in one step.

Interview me

Interviews the user one question at a time, each with a guessed answer and a confidence number, until the agent can predict the next replies, then confirms a statement of intent before any spec or code. Use when a request leaves who, why or success unclear.

Observability and instrumentation

Instruments a feature with structured JSON logs and request IDs, RED and USE metrics with bounded labels, OpenTelemetry traces and symptom-based alerts with runbooks, then tests the telemetry. Use when a feature, job or integration is headed for production.

Performance optimization

Optimizes by measurement: takes a baseline, fixes the one proven bottleneck, re-measures against noise and reverts neutral changes, with worked fixes for N+1 queries, indexes, caching and bundles. Use when something is slow or misses a performance budget.

Planning and task breakdown

Breaks a spec into small vertical-slice tasks with acceptance criteria, verification steps, sizes, dependency order and checkpoints, saved to tasks/todo.md or a named tracker. Use when a spec is ready but the work is too big to start in one go.

Security hardening

Hardens code with a short threat model, always, ask-first and never rules, and patterns for injection, XSS, access control, SSRF, secrets, dependencies, personal data and LLM output. Use when code handles user input, auth, sensitive data or outside services.

Shipping and launch

Prepares a production launch with a pre-launch checklist, a feature-flag rollout from 5% to 100%, thresholds for advancing or rolling back, an error-budget gate, a rollback plan and first-hour checks. Use when a feature or major change is about to ship.

Source-driven development

Grounds framework code in official docs: reads exact versions from the dependency file, fetches the relevant page, follows the documented API and cites full URLs, flagging anything unverified. Use when framework-specific code must match the version in use.

Spec-driven development

Writes a spec before code, surfacing assumptions and covering objective, commands, structure, style, tests and boundaries, then gates plan, tasks and code on approval, with capability maps and delta specs. Use when new work has vague requirements.

Test-driven development, red-green-refactor

Drives changes with red-green-refactor: finds the repository's own test commands, writes the failing test first, reproduces each bug with a test before fixing it, and sets rules on mocks and naming. Use when adding logic, fixing a bug or changing behavior.