Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Wcag Accessibility Audit

    1

    Accessibility (WCAG 2.2) audit skill for AI coding agents, plus a CLI: static source scan, real-browser runtime scan (Playwright + axe-core), verified findings, JSON and SARIF output.

    Free

    Security scanned
    wcag-accessibility-audit

    wcag-accessibility-audit

    Example session with this skill installed

    Audit src/components for accessibility and fix the critical issues.

    • Read your context and instructions
    • Compiled the wcag-accessibility-audit
    • Generated the image

    Scanned src/components (38 files) and http://localhost:3000 (home, cart).

    Confirmed: 3 critical, 2 serious · Rejected (false positives): 2 · Needs your input: 1

    Fixed

    • Buy button was a <div> (2.1.1, critical) — ProductCard.tsx:14 → <button type="button">
    • Quantity input had no label (1.3.1, critical) — ProductCard.tsx:15 → <label htmlFor>
    • Close icon button had no name (4.1.2, critical) — Modal.tsx:22 → aria-label="Close dialog"

    Needs your input

    • Hero image alt text (1.1.1) — Hero.tsx:8: what should the image communicate?

    Report: .wcag-audit/REPORT.md

    ProductCard.tsx

    PNG · 1536×1024

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    AI ships inaccessible UIs. This makes your agent fix them.

    Agent-generated interfaces are full of barriers: inputs with only a placeholder, icon buttons with no name, clickable <div>s a keyboard can't reach, low-contrast text. wcag-audit gives your agent a disciplined, five-phase accessibility audit against WCAG 2.2 AA, and has it fix what it finds.

    The phases

    1. Static scan of HTML, JSX/TSX, Vue, Svelte and Astro source (24 rules, each mapped to a WCAG success criterion), plus a manual review of the component layer the scanner can't see.
    2. Runtime scan in a real browser with axe-core, including Tab traversal for keyboard traps and invisible focus.
    3. Verification: the agent tries to disprove every finding by tracing props, parents and the rendered DOM. Only confirmed findings survive, so you get a short list of real bugs, not noise.
    4. Fix: minimal, native-HTML-first fixes in your framework's idiom, each explained (what was wrong, who it affected, what changed), then re-verified mechanically.
    5. Report: a validated findings.json, a readable REPORT.md, and an honest "not tested" list. It never claims "compliance".

    Works without a browser in static-only mode, and without the CLI in a manual mode. The same CLI runs in CI with SARIF output for GitHub code scanning.

    Runs locally. MIT-licensed, open source: github.com/mintwork/wcag-audit-skill.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.

    Automated checks catch roughly 30–50% of WCAG issues: alt-text quality, captions accuracy, reflow and most cognitive criteria still need human review, and the skill reports them as "not tested". It shows failures, not conformance. The runtime scan covers one state per URL; other states (open modals, error states) need extra URLs or a browser tool.

    Open docs

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed today

    • Free with an Agensi account

    Software
    Node.js 22+
    Chrome, Edge
    orPlaywright Chromium (runtime scan)
    @mintwork/wcag-audit-skillPlaywrightsvelte-check
    Optional
    A running dev server
    orURL (runtime phase)

    Permissions

    Allowed Hosts

    registry.npmjs.org (npx downloads the @mintwork/wcag-audit-skill CLI)
    localhost or URLs you name (runtime scan)

    File Scopes

    UI
    source
    files
    in
    your
    project
    (read;
    edited
    only
    the
    fix
    phase)
    .wcag-audit/
    at
    repository
    root
    (scan
    results
    and
    reports)

    Runs locally. It never scans systems you didn't name and never submits real forms.

    Listedtoday

    What's inside

    Frequently Asked Questions

    No, unless a finding is visual (contrast, focus ring, target size). For changes to visible UI or copy, the agent asks first.

    No tool can. It finds and fixes the failures automation and an agent can verify, and lists what still needs a human check, like screen-reader testing.

    Yes. The skill and the CLI are MIT-licensed open source at github.com/mintwork/wcag-audit-skill.

    Free