audit-accessibility

name

audit-accessibility

description

Deep standalone iOS accessibility audit. Dispatches the accessibility-engineer solo for the full treatment: VoiceOver semantics and reading order, Dynamic Type survival through AX5, color contrast, touch targets, Reduce Motion / Reduce Transparency, Switch Control, Voice Control, hearing and cognitive accessibility, WCAG 2.2 mapping. Triggers on "audit my app for accessibility", "accessibility audit", "VoiceOver review", "is my app accessible?", "WCAG check", "a11y pass", "test with Dynamic Type". Use this instead of review-ios-ui when accessibility is the whole ask -- it goes deeper than the 3-specialist review.

Audit Accessibility

Dispatch

This skill dispatches one specialist, solo, with the full accessibility reference set:

apple-ui-craft:accessibility-engineer
   - Scope: all files in the audit scope
   - Must-read: references/accessibility/ (all files), references/design/03 + 04 + 06,
     references/patterns/01 (env-key injectability), references/_scaffolding/version-floor-registry.md
   - Audits: VoiceOver (labels, traits, grouping, custom actions, rotor), Dynamic Type
     (through AX5, reflow at isAccessibilitySize), visual (contrast, Reduce Motion,
     Reduce Transparency, Smart Invert, Dim Flashing Lights), motor (44pt targets,
     Switch Control, Voice Control, drag alternatives), hearing (captions, Live
     Captions, Sound Recognition), cognitive (AssistiveAccess, plain language)

Solo dispatch goes deeper than the review-ios-ui a11y lane: the engineer runs its full 5-dimension audit (VoiceOver, Dynamic Type, visual, motor, cognitive & hearing -- every check table in agents/accessibility-engineer.md) per screen, maps every finding to a WCAG 2.2 success criterion with the correct level (2.5.8 Minimum AA 24px vs 2.5.5 Enhanced AAA 44px -- 44pt stays the Apple HIG hard rule), and produces the Settings matrix (what changes under each accessibility setting).

Scope determination

Arg	Meaning
empty / diff	Uncommitted + staged changes (default)
staged	Only staged files
<file> or <directory>	Specific target
all	Entire project (excluding generated code, Pods, build artifacts)

Output

• Accessibility verdict: INCLUSIVE / ADEQUATE / GAPS / EXCLUDING
• The evidence mode and which accessibility settings were actually exercised. A setting that was not turned on is reported not exercised, never OK -- a clean cell filled from a default-configuration screenshot is a false negative on the users this audit exists for (references/review/02-evidence-pipeline.md#configuration-coverage)
• Findings in the canonical format (references/review/01-finding-format.md), each carrying its WCAG: and Who is affected: lines
• Findings by screen, severity-tagged (CRITICAL = exclusion or WCAG A/AA violation)
• Per-finding: WCAG criterion + level, affected assistive technology, concrete SwiftUI rewrite
• Settings matrix: behavior under Reduce Motion, Reduce Transparency, Increase Contrast, Bold Text, AX text sizes, VoiceOver, Switch Control
• Top 3 priority actions

All findings are advisory. The user chooses what to apply.

Execution mode

The specialist this skill dispatches is pinned to Opus 5 (model: "opus") at dispatch -- the coding/review floor (owner directive 2026-07-24); the session conductor stays orchestrator-only. When the session model is already the strongest tier and the audit scope is small, the orchestrator may run the audit inline in the main context (foreground) instead of dispatching a separate agent, without weakening the accessibility-engineer's read-only guarantee.

Review ledger (write it, without asking)

At the end of every run, write .claude/apple-ui-craft/last-review.json in the REVIEWED repo, and read any existing one first to produce a delta report (NEW / RESOLVED / STILL OPEN / REGRESSED / IMPROVED). Schema, matching rules, and the two fields that stop a narrow run from erasing a wide one (dimensions and notAssessed): references/review/04-run-artifacts.md#review-ledger-schema-v1.

A missing or unreadable ledger is an empty prior run, never an error. A ledger from a different scope is not a prior run for this scope.

Ultracode conductor mode

When the harness announces ultracode, this skill runs conductor-executor per references/_scaffolding/conductor-dispatch-protocol.md -- read that file before the first executor dispatch; it owns the dispatch mechanics, the fan-out doctrine (executor teams scale to natural breadth; the session-model agent caps do not apply to them), the executor prompt contract, and the validation gate. Without ultracode, run the standard dispatch above unchanged.

Split of labor

Conductor (session model -- never delegated)	Conductor-selected executors (Sonnet 5 @ xhigh or Opus 5)
Scope decision, severity verdicts, WCAG-level grading, finding dedup, final report synthesis -- accessibility verdicts are always conductor-class	Per-screen a11y evidence collection against the engineer's 5-dimension check tables; VoiceOver label/trait/order inventory; contrast-pair computation sweeps; Dynamic Type breakpoint capture

Executor scoping (on top of the protocol's prompt contract)

• Reference set: absolute paths of references/accessibility/ + references/_scaffolding/version-floor-registry.md.
• Inline the severity scale and the relevant dimension tables from agents/accessibility-engineer.md -- its 5-dimension framework is the checklist executors collect evidence against.
• The accessibility-engineer specialist is pinned to Opus 5 at dispatch -- judgment reviewer, never a grunt executor.