Content
50%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill delivers genuinely valuable, executable material in its automated-testing and remediation-pattern sections, but buries it in a ~550-line monolith that re-enumerates the entire WCAG 2.2 checklist and explains concepts Claude already knows. Restructuring into an overview plus a references file, and adding an explicit audit workflow, would substantially improve it.
Suggestions
Move the per-principle WCAG checklists (Principles 1-4) into a references/ file (e.g. references/wcag-checklist.md) and keep only the high-impact 'Common Violations by Impact' summary and links in SKILL.md.
Delete or drastically shrink the 'Core Concepts' section (conformance levels, POUR definitions) — this is background knowledge Claude already has and costs tokens without adding capability.
Add an explicit sequenced audit workflow, e.g. 1. Scope the audit (target level A/AA), 2. Run automated scan (axe CLI), 3. Manual keyboard/screen-reader checks per checklist, 4. Triage findings by impact, 5. Apply remediation patterns and re-run the scan to verify fixes.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body spans ~550 lines, of which the 'Core Concepts' section (POUR principles, conformance levels, 'Perceivable: Can users perceive the content?') and the ~350-line enumeration of every WCAG success criterion ('1.4.3 Contrast (Minimum)', '2.4.11 Focus Not Obscured', etc.) restate reference knowledge Claude already has. Several padded sections duplicate the WCAG spec rather than adding what Claude doesn't know. | 2 / 5 |
Actionability | Mostly executable, copy-paste-ready guidance: the axe-core integration ('await page.addScriptTag({ path: require.resolve(\'axe-core\') })'), CLI commands ('npx @axe-core/cli https://example.com', 'npx pa11y'), and before/after remediation patterns for labels, contrast, and keyboard access. Minor gaps keep it below 5: the AccessibleDropdown example calls undefined methods (this.toggle(), this.focusNext()) and the Playwright snippet omits its imports/setup. | 4 / 5 |
Workflow Clarity | The checklist is systematically organized by principle and the layout implies a rough order (checklist -> automated testing -> remediation), but there is no explicit end-to-end audit sequence tying together scoping, automated scan, manual verification, and reporting. Checkpoints are implicit (e.g., no step for triaging 'Critical (Blockers)' vs 'Serious' findings or verifying fixes after remediation), matching the 'steps present but checkpoints missing or implicit' anchor. | 3 / 5 |
Progressive Disclosure | Section headers are clear ('## Audit Checklist', '## Automated Testing', '## Remediation Patterns'), but the skill is a single monolithic file with no bundle: the per-principle checklists (hundreds of lines, e.g. the full Perceivable/Operable/Understandable/Robust sections) are exactly the content that should live in a separate references file. This matches 'some structure but content that should be separate is inline'. | 3 / 5 |
Total | 12 / 20 Passed |