agentsclimarketplace

A11y deep

Skill mj-deving/pai-skills/skills/Frontend/Impeccable/a11y-deep

Curated, sanitized export of 21 agent-skill packages for Claude Code and Codex, gated by an automated publication audit (no secrets, no local paths).

Install
npx -y skills add mj-deving/pai-skills --skill a11y-deep

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Deep WCAG 2.2 AA accessibility audit using specialist agents. Use for public-facing pages, portfolio deploys, or when Impeccable/audit flags a11y issues. Slower than /audit but catches everything.

SKILL.md

4.1 KB, as published. Nobody here has run it

Deep Accessibility Audit

Full WCAG 2.2 AA compliance review using specialist agents from the accessibility-agents plugin. This is the deep dive — use it before shipping public-facing work.

When to Use

SituationUse /audit (quick)Use /a11y-deep (this)
Every buildYesNo
Before deploying portfolioYes first, thenYes
Complex forms or modalsMaybe enoughYes
Impeccable/audit flags a11y score < 3Already ranYes — dig deeper
Client/public-facing workYes first, thenYes
Internal dashboardsYesNo — overkill

Workflow

Step 1: Identify the target

Read the file(s) to audit. Accept:

  • A single HTML file path
  • A component directory
  • A URL (if Playwright MCP is available for live audit)

Step 2: Run specialist agents (parallel where possible)

The accessibility-agents plugin lives at: ${AGENT_HOME}/plugins/cache/community-access/accessibility-agents/3.2.0/agents/

Load the relevant specialist prompt files and apply their checklists to the target code. Run these checks yourself using the specialist's checklist — do NOT try to spawn them as subagents (the subagent_type is not registered).

Core specialists to consult (always):

SpecialistFileChecks
Contrast Mastercontrast-master.mdColor contrast ratios, focus indicators, dark mode, prefers-* media queries
ARIA Specialistaria-specialist.mdRoles, states, properties, widget patterns, landmark structure
Keyboard Navigatorkeyboard-navigator.mdTab order, focus management, skip links, keyboard traps
Alt Text & Headingsalt-text-headings.mdImage descriptions, heading hierarchy, landmark labels
Link Checkerlink-checker.mdAmbiguous link text ("click here", "learn more"), new-tab warnings

Conditional specialists (invoke when relevant):

SpecialistFileWhen
Forms Specialistforms-specialist.mdPage has <form>, <input>, <select>, <textarea>
Modal Specialistmodal-specialist.mdPage has modals, dialogs, drawers, popovers
Tables Data Specialisttables-data-specialist.mdPage has <table> elements
Live Region Controllerlive-region-controller.mdPage has dynamic content updates, toasts, notifications
Design System Auditordesign-system-auditor.mdAuditing a component library or design system

Step 3: Compile findings

Format as severity-rated report:

## Deep Accessibility Audit: [filename]

### Summary
- **WCAG Level:** [A / AA / Partial AA / Fails AA]
- **Critical Issues:** [count]
- **Major Issues:** [count]
- **Minor Issues:** [count]

### Critical (must fix before ship)
1. [WCAG criterion] — [issue] — [fix]

### Major (should fix)
1. [WCAG criterion] — [issue] — [fix]

### Minor (nice to fix)
1. [WCAG criterion] — [issue] — [fix]

### Passing
- [List what passes — gives confidence, not just problems]

Step 4: Fix or report

  • If invoked with "fix" in the request: apply fixes directly
  • If invoked as audit only: report findings, let user decide

What This Does NOT Cover

These specialists are NOT included (out of scope for web frontend):

  • PDF/Word/Excel/PowerPoint accessibility (use document-accessibility-wizard if needed)
  • Desktop app accessibility (NVDA, JAWS — irrelevant for web)
  • GitHub DevOps (25 agents for repo management — not a11y)
  • Mobile native (React Native, SwiftUI — separate concern)

Integration with Impeccable Pipeline

Quick gate:    /audit           → 5-dimension scored report (run every time)
Deep dive:     /a11y-deep       → full WCAG specialist review (run before deploy)
Production:    /harden          → cross-browser, edge cases, performance

The recommended flow in BrandedBuild Step 4 is: /audit first. If a11y score < 3, escalate to /a11y-deep.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.