accessibility-audit
Skill0 starsChanged 5 days agoMITView on GitHub
Checks a website's color contrast against WCAG AA and AAA (text, text on photos and gradients, placeholders, form field edges, icon buttons, links told apart only by color) in light and dark mode at desktop and phone width, lists very small text, gives each issue a priority and a passing color, and scores the site as a percentage for AA and AAA. Then offers to fix it. Use it whenever the user asks for an accessibility audit, a11y check, WCAG or contrast check, "is my text readable", "does it pass AA", "audyt dostępności", "sprawdź kontrast", or before launching a site.
- Installs packages
58 other projects use a version of this.
Adopted by
Most recently taken up by
- antick/skillstheir own version, first seen Nov 16, 2026
- magnusrodseth/dotfilestheir own version, first seen Nov 16, 2026
- Cure-Consulting-Group/ProductEngineeringSkillstheir own version, first seen Nov 12, 2026
- ishandutta2007/Awesome-Claude-Skillstheir own version, first seen Nov 8, 2026
- ainurdev/mystical-assistanttheir own version, first seen Nov 8, 2026
- mzeleniuk/ClaudeVueProjectTemplatetheir own version, first seen Nov 8, 2026
Used with
Projects with it also use design-system-audit, flaky-test-triage and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
jpoindexterSkilldesign-system-audit
Audit code for design system violations - grid spacing, tokens, existing components
Also in 4 projects that use this
SUDARSHANCHAUDHARISkillflaky-test-triage
Diagnoses inconsistent test results by separating product nondeterminism, automation defects, shared data, timing, infrastructure, environment, order, and concurrency effects.
Also in 3 projects that use this
tekumaraSkillapi-design-review
Reviews REST API design by inventorying routes, finding overlapping functionality, comparing payloads and pagination, and checking resource naming, URL structure, identifiers, errors, idempotency, and timestamps. Use when asked to review API routes, compare endpoints, find redundant or superseding operations, assess contract consistency, or run a REST API design review.
Also in 3 projects that use this
BufferoverflovvSkillobservability-setup
Add production observability to a FastMCP server: request IDs, slow-request logging, structured log fields, and optional Prometheus metrics. Use when: hardening logging for production, adding request tracing or correlation IDs, setting up slow-request alerts, adding a /metrics endpoint, or improving structured log output beyond the base middleware.
Also in 3 projects that use this
mmondoraSkillinfrastructure-as-code
Infrastructure as Code with Terraform and Pulumi. State management, modularity, drift detection, secrets handling. Use when provisioning cloud resources, writing Terraform/Pulumi modules, or managing infrastructure configuration.
Also in 3 projects that use this
OpenAISkilldesign-qa
Internal prototype QA helper. Use only after a Product Design prototype, URL-to-code build, or image-to-code build has a source visual target and a rendered implementation to compare before handoff. Do not use for broad UX critique, design critique, product audits, or flow reviews; route those user-facing requests to audit.
Also in 3 projects that use this
Other versions58 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 58
- me2resh/apexyard★ 542same name
- github/replica★ 275same name
- cuellarfr/design-skills★ 94same name
- adityareddy-dev/deskstudio★ 43same name
- launchapp-dev/animus-cli★ 39same name
- kumaran-is/claude-code-onboarding★ 35same name
- 5throck/ai-workspace-standards★ 25same name
- Indium-AI-Labs/indium-agentkit★ 18same name
- Ionate-Technologies/indium-agentkit★ 17same name
- ishandutta2007/Awesome-Claude-Skills★ 16same name
48 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- copy-edit-modeSkill
- design-cleanupSkill
- tweak-modeSkill
- website-auditSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 49 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 38
Read site-reports/accessibility.json.
- Very specific: line 57
Ask once which colors to use: the nearest passing shade (fix.color, keeps the look) or a color from their palette (fix.palette, a token …
- Distinctive: line 60
Shared tokens: if the color comes from a token used elsewhere, check where else it's used.
- Very specific: line 65
Links in text with no underline: add text-decoration: underline (and a small text-underline-offset) to links inside paragraphs.
- Distinctive: line 67
Small text: no rule forbids it, so only change it if they ask.
- Distinctive: line 72
Run the audit again and give the new AA and AAA numbers and anything still failing.
“Common” is measured: kept by at least 6 in 10 of the 49 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Accessibility audit
Renders every page in the browser and measures the colors people actually see. Writes a report page, then fixes what the user approves.
Step 1 - Find the site
- A URL the user gives, or their dev server (
npm run dev, then its URL). Start it if it isn't running. - A folder of plain HTML: pass the folder, the script serves it.
Step 2 - Run it
<skill-dir> is the folder this file is in. Needs Node 20+ and Chrome or Edge on the computer.
First time only, if <skill-dir>/scripts/node_modules doesn't exist:
npm install --prefix <skill-dir>/scriptsThen:
node <skill-dir>/scripts/a11y_audit.mjs --url http://localhost:3000 --out site-reports
node <skill-dir>/scripts/a11y_audit.mjs --root ./site --out site-reportsOptions: --pages /,/about checks only those, --max 20 caps how many pages it finds by following links, --light skips dark mode.
It takes one to three minutes. It prints a short summary as JSON.
Step 3 - Show the result
- Open the report for the user:
open site-reports/accessibility.html(Windows:start, Linux:xdg-open). - Read
site-reports/accessibility.json. Each issue haspriority(high, medium, low),title,ratio,need,level,fixandelements(withselectorandsource: the CSS rule that sets the color). - Reply in the user's language, short:
AA: 86% · AAA: 61%
Fix first
- Footer links: grey on black, 2.6:1 (needs 4.5:1). 14 places.
- ...
Fix next: 3 · Only fails AAA: 5 · Very small text: 2
Fix the "fix first" ones now? I'd use the nearest shade that passes, so the look barely changes.What the levels mean, if they ask: AA is the usual legal and industry bar. AAA is stricter and optional. A site meets a level only when every check passes; the percentage shows how close it is.
Step 4 - Fix what they approve
Ask once which colors to use: the nearest passing shade (fix.color, keeps the look) or a color from their palette (fix.palette, a token they already have). Default to the nearest shade.
- Change the color where
sourcesays it's set (source.file,source.selector, insidesource.mediaif given). In app projectssource.fileis built CSS: find the real rule by its selector, or the Tailwind class. - Shared tokens: if the color comes from a token used elsewhere, check where else it's used. Changing the token is cleanest, but it changes every use: say so and ask. Otherwise override it only where it fails.
- Dark mode issues (
elements[].scheme: "dark"): edit the dark theme's values, not the light ones. - Faded text (
fix.opacity): the text is see-through because ofopacityon it or a box around it (elements[].opacitySourceis that rule; Tailwindopacity-50). Changing the color alone keeps the fade and can make it worse. Raise the opacity tofix.opacity.needed%, or remove it and usefix.colorwhen given. - Text on a picture (
needsInput: true): ask which they prefer: a dark gradient behind the text, a solid shade, or moving the text off the picture. - No color reaches it (
needsInput: true, nofix.color): only a different background helps. Ask. - Links in text with no underline: add
text-decoration: underline(and a smalltext-underline-offset) to links inside paragraphs. No need to ask. - Form field edges: set the border to
fix.color. - Small text: no rule forbids it, so only change it if they ask.
- Change only colors (and underlines). Don't touch layout or copy.
Step 5 - Check again
Run the audit again and give the new AA and AAA numbers and anything still failing.
Notes
- Reports go to
site-reports/. Suggest adding it to.gitignore. - Dark mode: it checks dark mode when the site follows the system setting. If the site switches dark mode with a class (a theme toggle), only light is checked; the JSON says so in
darkMode. Tell the user. - Hidden, collapsed and decorative content (
inert,aria-hidden) is skipped. Text that fades in on scroll is measured after scrolling to it. - It checks colors and text size. It does not replace testing with a keyboard and a screen reader; say so if they ask whether the site is "accessible".
- Text on photos is measured from the pixels under the text. Moving pictures (video, WebGL) are measured as they were at that moment.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
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.

