a11y-core
Shared contract for the Accessibility Agents skills - dispatch, findings schema, report rules. Read by skills, never dispatched…
Open Source Accessibility Framework Testing phase: automated, keyboard, screen reader, manual and docs checks.
Claude Code
$ npx -y skills add Community-Access/accessibility-agents --skill osaf-testing --agent claude-code
Codex
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a codex -y
Cursor
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a cursor -y
Gemini CLI
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a gemini-cli -y
GitHub Copilot
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a github-copilot -y
OpenCode
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a opencode -y
Windsurf
$ npx skills add Community-Access/accessibility-agents --skill osaf-testing -a windsurf -y
name: osaf-testing description: "Open Source Accessibility Framework Testing phase: automated, keyboard, screen reader, manual and docs checks." license: MIT disable-model-invocation: true metadata: tier: specialist domain: github output: artifact effort: medium title: "OSAF Testing Phase" osaf-phase: testing osaf-commit: "b2ea31a2e38cbab3e774e4142575b78876b9b53f" osaf-digest: "b1fd16cfeb8b" osaf-source: "https://raw.githubusercontent.com/open-source-accessibility/Open-Source-Accessibility-Framework/main/framework/phases/testing-phase.md"
You implement the Testing phase of the Open Source Accessibility Framework in one repository: all six actions, or those the request names. You set up the checks, write the test procedures, record what was actually run, and return the work for review. Facts and the action index: `kb-open-source-accessibility-framework`.
1. **Scope.** Every action in phase order, or only those named by number, title or ID. If none is named, ask before changing files. Report an unknown action and list the valid ones. 2. **Bundled requirements.** The Definitions of Done below are the minimum, from the phase document at `metadata.osaf-commit`, and work offline. Read an action's reference only when it is in scope. 3. **Online check, only when asked.** Fetch `metadata.osaf-source`; if the first 12 hex characters of its SHA-256 differ from `metadata.osaf-digest`, report the change and ask before using it. A failed fetch never blocks the work. 4. **Inspect before editing.** Per item: met (cite file and line), needs a change, or needs maintainer input, external configuration or a person. 5. **Edit in phase order**, keep repository conventions, extend `ACCESSIBILITY.md` rather than replace it. 6. **No test without evidence.** Never claim a manual, keyboard, screen reader, visual or assistive technology test was performed without runtime evidence. A simulation or a static code review is not a test run. 7. **Record the environment** of every test that ran (operating system, browser, assistive technology, versions), and separate tests that ran from procedures that still need a person. 8. **Verify and present.** Per action: met, unmet, recurring work, follow-up. Present for review before anything is committed or published. 9. **No compliance claims.** No automated tool finds every barrier, and the framework is for improvement, not certification.
Work another skill does better goes back as a hand-off, named in each action reference; `github-hub` passes web testing to `accessibility-lead` or `web-accessibility-wizard` and documentation to `markdown-a11y-assistant`.
1. Add at least one automated accessibility check - `add-at-least-one-automated-accessibility-check`, Initial setup + ongoing - `references/action-01.md`
2. Perform a keyboard-only smoke test for core flows - `perform-a-keyboard-only-smoke-test-for-core-flows`, Ongoing - `references/action-02.md`
keyboard; can be invoked by keyboard; provide a visible focus indicator; provide a logical focus order; have no keyboard traps.
3. Perform a screen reader spot check for core flows - `perform-a-screen-reader-spot-check-for-core-flows`, Ongoing - `references/action-03.md`
4. Perform manual accessibility checks - `perform-manual-accessibility-checks`, Ongoing - `references/action-04.md`
language, plain language, zoom and reflow, and color contrast and usage.
5. Perform accessibility checks for documentation - `perform-accessibility-checks-for-documentation`, Ongoing - `references/action-05.md`
tables, and code blocks.
6. Document supported environments and known limitations - `document-supported-environments-and-known-limitations`, Initial setup + ongoing - `references/action-06.md`
browsers, input methods, and assistive technologies.
users, with workarounds, equivalent access, and links to tracked issues where available.
"Fixed or tracked" means each barrier found is fixed or has an issue with the accessibility label; `issue-tracker` files them once approved.
Write the files, then report: files and what each change does; each Definition of Done item as met, unmet or needs a person, with evidence; every test as ran (with environment and result) or not yet run; hand-offs (skill, task, scope); tracking issue lines that can now be checked (never edit it). Shape: `examples/sample-output.md`. License: `references/LICENSE-OSAF.md`.
Shared rules, dispatch contract and schemas: `skills/a11y-core/SKILL.md`. Authoritative specifications for this skill: `skills/a11y-core/references/sources.md`.
WCAG 2.2 AA enforcement for agentic coding, as a set of Agent Skills. One package, read natively by Claude Code, Codex, GitHub Copilot, Gemini CLI and Antigravity, with no per-client copies. Models forget accessibility while generating code.
Shared contract for the Accessibility Agents skills - dispatch, findings schema, report rules. Read by skills, never dispatched…
Build accessibility scanners, rule engines, parsers and report generators.
Web UI accessibility lead. Use before writing or changing HTML, JSX, TSX, Vue, Svelte, CSS or templates. Picks specialists and…
Compare audits across commits to find new, fixed and regressed issues.
Generate a W3C or EU model accessibility statement from audit results.
GitHub Actions: workflow runs, logs, re-runs and CI failure triage.
Produce the "Playwright 1.xx API Review" artifact for a range of commits - the public API additions and the breaking or behavior…
Write, refine, run, and QA promptfoo evaluation suites: promptfooconfig.yaml, prompts, providers, vars, tests, assertions,…
DeepEval evaluation workflow for AI agents and LLM applications. TRIGGER when the user wants to evaluate or improve an AI agent,…