Dev & Engineering swiftuiiosaccessibilitywcag-22code-auditvoiceoverstatic-analysis

SwiftUI WCAG Accessibility Auditor Skill

Audits SwiftUI iOS code against WCAG 2.2 A/AA and returns priority-grouped, patch-ready fixes.

FollowSkills review · FSRS-2.0
Recommended
58/ 100 5-point scale 2.9 / 5
1 2 3 4 5 6
1Trust20 / 25 · 4.0/5

Read-only static code audit skill: does not run apps, modify files, access networks, or exfiltrate data; least-privilege posture is clear, and mandatory 'Needs user verification' flags prevent overclaiming. Deducted for: no explicit confirmation/rollback disclosure, unverified publisher identity, and missing license metadata, so full marks are not justified.

2Reliability9 / 20 · 2.3/5

High instruction self-consistency: load order, status vocabulary, strict output template, and static-evidence limits align across files with no contradictions; rg commands are concrete. Deducted for: static review cannot execute key paths, actual existence/consistency of referenced files is unverified, and failure feedback on abnormal input is only inferable from docs; capped by the static ceiling.

3Adaptability12 / 15 · 4.0/5

Clear audience and scope: native iOS SwiftUI audits, explicit non-fit exclusions (macOS, watchOS, web, UIKit-dominant), and routing rules to nearby skills; trigger semantics via example requests are precise. Deducted for: no Chinese-language support declared and uncertainty from reliance on the WCAG2Mobile draft.

4Convention7 / 15 · 2.3/5

Good layering: SKILL.md plus six references with progressive disclosure, TOCs, non-goals, priority model, and a manual checklist. Deducted for: unknown license, no versioning/changelog, undeclared maintenance ownership, and the wcag2mobile reference exposing the author's local absolute path (/Users/romanmirzoyan/...), indicating incomplete provenance hygiene.

5Effectiveness6 / 15 · 2.0/5

README shows a plausible, directly usable end-to-end audit report (P1 findings, WCAG mapping, patch-ready snippets, follow-up checks), so the core task is structurally achievable. Deducted for: the sample output is not independently executed-verified, patch correctness cannot be statically confirmed, and comparative-benefit evidence is thin; capped at 7 by static calibration.

6Verifiability4 / 10 · 2.0/5

Evidence discipline is strong: findings require file path, symbol, and WCAG SC mapping; normative vs informative sources are separated and uncertainty is flagged. Deducted for: the primary source (wcag_doc.md) points to a local author path and cannot be independently verified; no third-party execution evidence, tests, or CI coverage of key paths; capped at 5 with limited evidence-type diversity.

Evidence confidence:Low Reviewed Sep 10, 2026 Reviewed revision 8214fbf61e3a
Before you use it
  • License metadata is missing; confirm licensing before enterprise use.
  • Reference files cite the author's local absolute path; the WCAG2Mobile source cannot be independently verified, leaving provenance incomplete.
  • This was a static review without execution: audit quality and patch correctness are inferred from docs and the README sample only; pilot-test on a real codebase first.
  • No versioning, changelog, or declared maintenance ownership; long-term availability is uncertain.
  • No Chinese-language support declared; output language must be confirmed by the user.
  • WCAG2Mobile is an informative draft (many sections placeholder/WIP); the skill labels this correctly, but users should not treat it as a normative conformance standard.
Review evidence [1][2][3][4][5][6][7][8]
See the full review method →

What does this skill do, and when should you use it?

This Agent Skill guides an AI to audit SwiftUI iOS screens for accessibility against WCAG 2.2 Level A/AA and its mobile interpretation, using code-only review. It never runs the app: each finding cites evidence, a WCAG success criterion, a confidence level, and a minimal Swift code patch. Anything indeterminate from source is flagged as a 'needs user verification' follow-up check. Reports follow a strict, concise Markdown format grouped by P0/P1/P2 priority. The repo ships six reference files covering the audit workflow, checklist, WCAG2Mobile interpretation, SwiftUI API examples, remediation strategy, and a manual validation checklist.

Reads SwiftUI view source in your repo and searches with rg for accessibility modifiers, gestures, and Dynamic Type patterns; walks a WCAG 2.2 checklist recording evidence and statuses per criterion; produces a Markdown report with findings grouped by priority (P0/P1/P2), each containing file/line location, WCAG SC mapping, confidence note, and an embedded patch-ready Swift snippet; and outputs user follow-up checks only for code-indeterminate items.

  1. An iOS developer pre-release auditing a SwiftUI checkout flow for a prioritized WCAG 2.2 findings list
  2. A team reviewing a SwiftUI login + OTP flow for accessible authentication conformance (SC 3.3.8)
  3. A maintainer of a gesture-driven task list checking VoiceOver/Switch Control role and semantic exposure
  4. A developer whose toasts rely only on visual/haptic feedback needing a 4.1.3 Status Messages fix for VoiceOver announcements
  5. A compliance-minded team needing every iOS fix mapped to a specific WCAG criterion for traceability

What are this skill's strengths and limitations?

Pros
  • Every finding maps to a specific WCAG 2.2 success criterion, including WCAG2Mobile mobile interpretation, for compliance traceability
  • Outputs minimal, patch-ready Swift code changes rather than abstract advice
  • Honestly labels confidence and routes code-indeterminate items to explicit human verification instead of guessing
  • Strict, consistent report format (priority groups, inline snippets) that drops straight into code review
Limitations
  • SwiftUI-native only; features delegating core behavior to UIKit are redirected to the companion mobile-accessibility-audit skill
  • Code-only audit — contrast values and actual VoiceOver announcement timing still require manual verification
  • No coverage of macOS, watchOS, or cross-platform frameworks
  • Repository license is unknown; confirm with the author before commercial use
  • README shows no automated tests or independent validation evidence

How do you install this skill?

Clone https://github.com/ramzesenok/iOS-Accessibility-Audit-Skill and place the skill folder (SKILL.md plus references/) into your Agent Skills directory. Exact install location and client configuration are not documented in the source; follow your platform's Agent Skills docs.

How do you use this skill?

Trigger it in an Agent Skills-compatible client with a prompt like: 'Use $swiftui-wcag-accessibility-auditor to audit this SwiftUI checkout feature against WCAG 2.2 and return prioritized findings with patch-ready fixes.' Point it at the SwiftUI views or feature files to audit; it returns a report grouped by P0/P1/P2 with embedded code snippets, plus user follow-up checks where behavior cannot be verified from code.

How does this skill compare with similar options?

SKILL.md itself names two adjacent skills: mobile-accessibility-audit (for features that may include UIKit, or standards-first coverage) and swiftui-accessibility-auditor (fast SwiftUI-only heuristic review without WCAG traceability). This skill is the merged, dual-mode option.

FAQ

Does it fix my code automatically?
No. It returns patch-ready snippets and suggestions for you to apply; the README is explicit that it provides an audit, not fixes.
How reliable are the findings?
It is a static, code-only review and never runs the app. Each finding carries a confidence label, and items indeterminate from code (e.g., actual VoiceOver announcements) are explicitly marked as needing user verification with concrete manual steps.
My app mixes in UIKit — what then?
UIViewRepresentable/UIViewControllerRepresentable bridges are audited only as far as they affect the SwiftUI feature. If core behavior is delegated to UIKit, the skill itself recommends switching to the mobile-accessibility-audit skill.
Does it need network access or special tools?
The search commands in the skill body rely on rg (ripgrep) and local file reads. The audit is local static analysis — no network calls or external services are required.

Related skills