SkillVaultskills Browse all 500 skills

Frontend · Version 1.5.0 · Reviewed 2026-08-02

Accessibility Review Agent

Improve keyboard navigation and semantic structure with evidence, explicit trade-offs, and a verification plan.

4 method steps 7 documented failure modes 5 diagnostic checks 7 quality gates

Reviews interfaces for keyboard access, semantics, contrast, and assistive-technology behavior.

₹99 one-time

Get this skill archive

What this skill helps you do

  • Keyboard navigation
  • Semantic structure
  • Contrast checking

How Accessibility Review Agent works

You provide

Component code, field metrics, and the failing interaction

It inspects

Render triggers and layout stability for keyboard navigation

It decides

A semantic structure fix targeting the measured vital

You verify

Field Core Web Vitals and keyboard traversal re-checked

What it checks first

Accessibility Review Skill reviews interfaces for keyboard access, semantics, contrast, and assistive-technology behavior. Use it when the work involves Keyboard navigation, Semantic structure, Contrast checking.

  1. Whether re-renders come from changed props, changed context, or a new object identity created during render.
  2. Which Core Web Vital is failing, since LCP, INP, and CLS have completely different causes and fixes.
  3. Whether state lives at the right level, because state placed too high re-renders subtrees that never read it.
  4. Effect dependency arrays that lie, either omitting a dependency or including an unstable one.
  5. Bundle composition: whether a single dependency dominates the critical path.

Failure modes it recognizes

  • An inline object or arrow function in props defeating memoization on every render.
  • A `useEffect` that sets state derived from props, causing a double render and occasional flicker.
  • Layout shift from images and embeds without reserved dimensions, damaging CLS after content loads.
  • A long task on the main thread blocking interaction response and inflating INP.
  • Stale closure capturing an old value inside an interval or subscription callback.
  • Hydration mismatch from rendering time, randomness, or browser-only APIs during server render.
  • Focus lost after a route change, leaving keyboard and screen-reader users stranded.

Answers it will reject

  • Wrapping everything in `memo` and `useCallback`, which adds comparison cost without removing the identity churn.
  • Fixing a race by adding a timeout, which reorders the symptom instead of the cause.
  • Using `aria-label` to patch a control that should have been a native element with real semantics.
  • Measuring performance in development mode, where the framework runs extra work that does not ship.

Decision rules it applies

  • Move state down or split context before reaching for memoization.
  • Derive during render instead of synchronizing with an effect; effects are for external systems.
  • Reserve space for anything that loads asynchronously to protect layout stability.
  • Prefer native semantic elements; ARIA is a correction layer, not a foundation.

Evidence it asks for

  • Profile with the framework profiler to attribute renders to a specific trigger.
  • Collect field Core Web Vitals rather than lab scores, since lab conditions hide real-device behavior.
  • Test keyboard-only navigation and screen-reader output for any interactive change.

The method inside

  1. Map the artifact, actors, boundaries, and invariants relevant to keyboard navigation.
  2. Trace concrete failure or abuse paths for semantic structure; do not report checklist items without a mechanism.
  3. Prioritize contrast checking findings by impact, likelihood, confidence, and cost of correction.
  4. Recommend the smallest defensible change, then define how an independent reviewer can verify it.

Deliverables

  • Keyboard navigation assessment
  • Semantic structure decision and action plan
  • Contrast checking verification checklist

Evidence requirements

  • Interface code, rendered behavior, and user journey
  • Browser/device matrix and accessibility tree
  • Performance and usability observations

Quality gates

  • Every material claim traces to supplied evidence or is labeled as a hypothesis.
  • The response follows the declared deliverable contract.
  • No execution, access, measurement, or verification is invented.
  • Secrets and personal data are redacted rather than repeated.
  • The user receives a concrete independent verification step.
  • The relevant failure modes in this domain were considered rather than only the reported symptom.
  • No listed anti-pattern was recommended as a solution.

Example task

Input

Review this modal component for accessibility problems before it ships to customers who use screen readers.

Expected output

Focus is never moved into the dialog and is not trapped, so keyboard users stay on the background page while the visual state says otherwise. The close control is also a div without a role, which means it is unreachable by keyboard and unannounced by assistive technology...

Boundaries and compatibility

Ideal for

  • Keyboard navigation: produce a decision or artifact grounded in supplied evidence.
  • Semantic structure: produce a decision or artifact grounded in supplied evidence.
  • Contrast checking: produce a decision or artifact grounded in supplied evidence.

Out of scope

  • Treating automated accessibility scans as complete
  • Changing visual style without preserving behavior

Agent compatibility

  • GitHub Copilot custom agents
  • Claude Agent Skills / SKILL.md
  • Any instruction-following chat model

Tool policy: Advisory by default. No tools are assumed. If the host provides tools, use read-only evidence gathering unless the user explicitly approves a scoped write or execution action.