a11y-checklist
Skill2 starsChanged 6 months agoMITView on GitHub
WCAG 2.2 AA quick reference — semantics, keyboard, names, states, contrast, motion, errors. Use when building or reviewing UI components.
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use this as a walk-through, top to bottom.
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Accessibility checklist (WCAG 2.2 AA)
Use this as a walk-through, top to bottom. If you can't answer "yes" or "N/A" for a line, fix it.
Semantics
- The element matches the intent.
<button>for actions.<a>for navigation. Lists use<ul>/<ol>. Landmarks use<main>/<nav>/<header>/<footer>/<aside>. - Headings form a valid outline — one
<h1>, no skipped levels. - Forms have
<label>s (oraria-labelledby). Placeholder is not a label.
Keyboard
- Every interactive element is reachable by
Tab. - Focus order matches visual order.
- No focus traps. Modals trap focus only while open and restore on close.
- Focus is visible (outline, ring, or theme equivalent). Don't remove
:focus-visiblewithout replacement. - Keyboard shortcuts don't collide with screen-reader virtual cursor.
Names & descriptions
- Every interactive control has an accessible name. Icon-only buttons:
aria-label. - Images: meaningful
alt; decorativealt="". - Form errors are programmatically associated via
aria-describedby.
States
aria-expanded,aria-selected,aria-pressed,aria-checked,aria-invalidreflect the current state and update as it changes.- Loading states have
aria-busyor a live region announcement. - Async status changes are announced in a
role="status"orrole="alert"live region (usealertsparingly — it's interruptive).
Contrast
- Normal text ≥ 4.5:1 vs background.
- Large text (≥ 18pt or 14pt bold) ≥ 3:1.
- UI components (form borders, focus rings, icon glyphs) ≥ 3:1.
- Don't rely on color alone to convey meaning — error state must have an icon or text too.
Motion
- Respect
prefers-reduced-motion. No auto-animating carousels, parallax, or transitions without the opt-out. - Videos under 5s ok; longer needs pause/stop controls.
- No flashing over 3 times per second in any region larger than ~25% of the viewport.
Structure & navigation
- Page has a
<main>landmark and a sensiblelangon<html>. - Skip-to-content link at the top (or equivalent landmark structure).
- Page title is descriptive and unique per route.
Errors & status
- Inline errors say what was wrong and how to fix ("Email must include an @").
- Errors appear near the input, are announced to AT, and don't rely solely on color.
- Required fields are marked both visually and programmatically (
aria-requiredorrequired).
Testing
- Keyboard-only pass: unplug the mouse, try the flow.
- Screen-reader pass (VoiceOver / NVDA) on critical flows.
- axe or similar linter in CI — treat findings as bugs, not noise.
Smells
role="button"on a<div>— use<button>insteadaria-hiddenon interactive content — almost always wrong- Tabindex > 0 — almost always wrong
- "Click here" link text — unhelpful out of context
Success criteria quick map
Common SCs you'll cite: 1.1.1 (non-text alt), 1.3.1 (info and relationships), 1.4.3 (contrast), 2.1.1 (keyboard), 2.4.3 (focus order), 2.4.7 (focus visible), 3.3.1 (error identification), 4.1.2 (name, role, value), 4.1.3 (status messages).
In this repository10 other files it gives its agents.
- api-designSkill
- context-engineeringSkill
- debug-playbookSkill
- migration-patternsSkill
- observability-checklistSkill
- perf-checklistSkill
- pr-review-rubricSkill
- security-checklistSkill
- test-patternsSkill
- token-economySkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

