Content
72%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is well-structured with concrete, executable a11y test commands and a clean single-level reference to the standards file. Its main weaknesses are redundancy across the manual-check, requirements, and fix sections and the absence of an explicit test→fix→re-run feedback loop.
Suggestions
Consolidate 'Common Accessibility Requirements', 'Manual Checks', and 'Fix Issues' into a single checklist to remove the repeated ARIA/focus/contrast items and tighten token usage.
Add an explicit feedback loop after automated testing, e.g. 'Review axe violations → fix in component → re-run `npm run test:a11y -- <component>` until clean'.
Drop generally-known tool enumeration (NVDA/JAWS/VoiceOver, the 4.5:1 contrast ratio) unless it adds non-obvious project-specific guidance.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient with concrete commands, but redundant — ARIA/focus/contrast recur across 'Common Accessibility Requirements', 'Manual Checks', and 'Fix Issues', and known concepts (NVDA/JAWS/VoiceOver, 4.5:1 contrast) are restated, so not every token earns its place. | 2 / 3 |
Actionability | Provides fully executable, copy-paste-ready commands ('npm run test:a11y', 'npm run test:a11y -- tabs button card') plus specifics on how component-name matching works, matching the score-3 anchor. | 3 / 3 |
Workflow Clarity | Steps are numbered 1–5 with a clear sequence, but there is no explicit validate/fix/re-run feedback loop (e.g., 'review axe violations, fix, re-run') — checkpoints are implicit rather than stated. | 2 / 3 |
Progressive Disclosure | A single, clearly-signaled one-level reference ('See .cursor/rules/accessibility-standards.md for details') with no nested references, and the body is organized into well-labeled sections appropriate for a single-purpose skill. | 3 / 3 |
Total | 10 / 12 Passed |