agentleFS
KyaniteLabsKyaniteLabs/tastecheck/skills/a11y-pass/SKILL.md

a11y-pass

Skill12 starsChanged 12 days agoMITView on GitHub

WCAG 2.2 AA fix pass for web UI. Use before shipping, or for requests about accessibility, a11y, keyboard navigation, screen readers, contrast, labels, focus, landmarks, target size, reduced motion, or ARIA.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write one path before opening a checker: entry → action → result/recovery.

    line 16
  2. Distinctive:

    Start with native HTML. Use ARIA only for relationships or live behavior native elements cannot supply.

    line 27
  3. Distinctive:

    Read the region with a screen reader or inspect computed name, role, value, state, and description.

    line 36
  4. Distinctive:

    Use this row format for every finding:

    line 48
  5. Very specific:

    Stop / handoff: Stop when the user path or target artifact is unavailable.

    line 85

“Common” is measured: kept by at least 6 in 10 of the 4 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Accessibility Pass (WCAG 2.2 AA)

Accessibility means finishing a real task without an avoidable barrier. Repair the failing path; prove it in the rendered interface.

Triage one rendered user path

Write one path before opening a checker: entry → action → result/recovery. Name outcome, affected UI, and smallest viewport or assistive setup that exposes failure.

Path pointInspectRepair when it fails
Entrylandmark, page title, reading order, skip linkgive the page a named, reachable start
Actionvisible focus, native semantics, keyboard order, target sizerepair the control before adding ARIA
Changeerrors, loading, dialogs, live updates, motionannounce the meaningful change without stealing focus
Recoveryerror copy, focus destination, retained input, retryreturn the person to a specific next action
Layout200% and 400% zoom, reflow, narrow viewportremove horizontal loss and clipped controls

Start with native HTML. Use ARIA only for relationships or live behavior native elements cannot supply. A role, audit score, or automated pass never proves the path.

Prove the path in the interface

  1. Tab forward/back from browser chrome through completion; record order, visible indicator, and traps.
  2. Trigger failure by keyboard; confirm purposeful focus, programmatic error association, and retained data.
  3. Read the region with a screen reader or inspect computed name, role, value, state, and description; confirm one announcement.
  4. Measure the rendered foreground/background pair; test reflow at 200% and 400%, including meaningful text over imagery and disabled-looking controls.
  5. Repeat animated interactions with reduced motion; the static path still exposes progress and completion.

A screenshot supports a claim but cannot prove keyboard order or announcements. Automated audits find candidates, not path clearance.

Report a repair ledger

Use this row format for every finding:

Path / criterionObserved failureRepairEvidenceResidual risk
Submit → invalid emailfocus lands on an unnamed summary; inline error is silentmove focus to named summary, link field error with aria-describedby, retain valuekeyboard trace; computed accessibility tree; 400% capturescreen-reader wording needs product review

Separate confirmed defect from design concern. State browser, assistive technology or inspection method, viewport/zoom, and exact reproduction step. If the artifact is absent, stop and ask for the rendered path; never invent evidence.

Delivery/handoff

Prioritize task blockers, frequent-path defects, then polish. Send component lifecycle or form-state questions to the owning skill when the defect is state design, not WCAG repair.

  • Audit method
  • Keyboard and focus helper
  • Decision records

Ship check

  • Named path has keyboard, semantic, zoom/reflow, and dynamic-state evidence.
  • Every finding names control/region, repair, and verification result.
  • Automated checks are discovery, not sign-off.
  • Focus, announcements, errors, and reduced-motion behavior have an owner.

Provenance — standard, not opinion

Grounded in the W3C Web Content Accessibility Guidelines (WCAG) 2.2, Level AA — the conformance target named throughout this skill (keyboard, focus, contrast, reflow, target size, reduced motion, error identification, status announcements). Independent Kyanite Labs synthesis of the standard's success criteria, credited — not copied. WCAG 2.2 is a W3C Recommendation; its success criteria are factual conformance requirements. Hand cognitive-load concerns to cognitive-a11y (COGA).

<!-- contract:v1:start -->

Contract (generated)

Canonical detail: contract.json.

  • Route: A rendered interface needs WCAG 2.2 AA repair findings verified by keyboard, browser, or measurement.; avoid: The request is only cognitive load or multilingual expansion without an accessibility defect.
  • Exclude: Do not self-attest a fix without browser or source evidence. (+1 in contract.json)
  • Stop / handoff: Stop when the user path or target artifact is unavailable. (+1 in contract.json); receives [component-states, form-ux, responsive-layout, theming, art-direction, cognitive-a11y, color-system, data-viz, deslop-ui, empty-states, i18n-ready, improve-existing-website, micro-motion, web-typography] -> sends [tastecheck-pass]
  • Output: prioritized WCAG repair ledger with measured verification
  • Evidence: table_with_evidence with status, reason, remediation, evidence, provenance.
<!-- contract:v1:end -->

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.