A11y Skills Collection
Over 30 accessibility skills, one per topic: forms, keyboard, contrast, ARIA, CI, manual testing and axe rules.
- 60 ★ on the repository
- Last commit
- AGPL-3.0
Minimal access
Why this level
- The skills consist of text rules and links to guides and do not run code or reach the network themselves.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/a11y-skills-collectionDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skills consist of text rules and links to guides and do not run code or reach the network themselves.
This is third-party code. Review the repository files before installing.
What it does
Each skill compresses a separate best practice guide from `examples/` into rules an agent follows. Topics include forms, keyboard, color contrast, ARIA live regions, anchor links, audio and video, charts, maps, Mermaid diagrams, navigation, plain language, print, light and dark mode, progressive enhancement, bug reports, manual testing and CI pipelines. The canonical source stays the file in `examples/` and a skill is its digest. A separate `axe-rules` skill explains axe-core rules, while `ai-workflows/planner` and `critic` handle planning and critique. The skills come with check sets in `evals/`, and advice on choosing between AI audits and scanners lives in `AI_AUDIT_GUIDE.md`.
Who it is for. Developers and accessibility auditors who need narrow skills per topic rather than one general checklist.
Good fit when
- A registration form needs to meet WCAG: labels, errors, focus order.
- Menus and modals are built to work from the keyboard alone.
- ARIA live regions announce updates but screen readers stay silent and the cause must be found.
- Accessibility checks by axe rules must be added to a CI pipeline.
Not a fit when
- A scanner is needed: skills guide the agent while axe automated checks are run separately.
- A legal conformance opinion is required: people make that assessment.
- The project is not web-based: topics are built around sites and WCAG.
Example request
Check the sign-up form with the forms skill: labels, error messages, focus order, and list what to fix.Limitations
The skills are derived from the guides in `examples/`, and on any mismatch the guide wins. Some skills describe practices that cannot be checked automatically: they need manual testing with a screen reader. The AGPL-3.0 license requires publishing changes when distributing network services built on the code. Installing with `npx skills add` puts skills into the local `skills/` folder of the project. For Claude Code there is a variant with `.skill` archives in `~/.claude/skills/`, and the page notes syncing with another repository.
How to disable. Delete the skill folders from the project `skills/` or the `.skill` files from `~/.claude/skills/`. Other project files stay untouched.
FAQ
How do I install a single skill?
With `npx skills add mgifford/accessibility-skills --skill forms`. Several skills are added by repeating `--skill`, and the command without that flag installs all.
How do I install globally in Claude Code?
Download a `.skill` file from the repository and place it in `~/.claude/skills/`, for example `curl -L <url>/light-dark-mode.skill -o ~/.claude/skills/light-dark-mode.skill`. Claude Code loads them on startup.
How does a skill differ from AGENTS.md?
The `AGENTS.md` or `ACCESSIBILITY.md` approach lives in the separate repository mgifford/ACCESSIBILITY.md. Here each topic is a self-contained skill.
When do I use an AI audit and when axe?
The `AI_AUDIT_GUIDE.md` document covers it: skills suit meaning and design review, while scanners such as axe catch formal violations.
Are there quality checks for the skills?
Yes, check sets in the `evals/` directory per topic, for example `forms` and `keyboard`, plus a skill sync check in CI.
Discussion
No questions or reviews yet.
Related
SkillsA six-skill accessibility plugin: page review, WCAG 2.2 audit with Playwright, audit planning and an improvement roadmap.
A library of 169 QA skills per language: test design, API, UI, regression, load, security and LLM testing.
Validates Agent Skills packages: structure, links, token weight and instruction quality, with optional LLM-as-judge scoring
Eleven skills to verify AI-written code: change blast radius, defect scan, review report, test cases and browser flow replay
Fifty testing skills for Claude Code, Codex and Cursor: Playwright, strategy, API, visual and load checks.
The evidence-code-review skill for Codex and Claude Code: evidence-based change review with MUST, SHOULD, BETTER and NITS levels
README in short
The project page gives a table of skills with their sources and install options: through `npx skills`, for Claude Code, for Claude.ai projects and for other agents.