Content
35%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill reads more like a high-level process outline than actionable guidance. It lacks any concrete tools, commands, code examples, or specific WCAG criteria references that would make it useful for actually performing an accessibility audit. The workflow structure is reasonable but needs validation checkpoints and the content relies heavily on an external playbook without providing enough standalone value.
Suggestions
Add concrete tool recommendations and executable commands (e.g., `npx axe-core`, `lighthouse --accessibility`, specific browser DevTools steps) to make the instructions actionable.
Include at least one concrete example of a finding mapped to a WCAG criterion with severity and remediation code (e.g., missing alt text → WCAG 1.1.1 → specific HTML fix).
Add explicit validation checkpoints, such as 'Verify zero critical violations in axe scan before proceeding to manual checks' and define pass/fail criteria for the re-test step.
Remove the redundant 'Context' section and the opening paragraph that restates the description to improve conciseness.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill has some unnecessary verbosity—the opening paragraph restates the YAML description, the 'Context' section largely repeats the 'Use this skill when' section, and 'Do not use this skill when' adds marginal value. However, it's not egregiously padded. | 2 / 3 |
Actionability | The instructions are entirely abstract and vague—'Run automated scans,' 'Perform manual checks,' 'Map findings to WCAG criteria' without specifying any concrete tools, commands, code snippets, or specific techniques. There is nothing executable or copy-paste ready. | 1 / 3 |
Workflow Clarity | There is a sequential list of steps (confirm scope → scan → manual checks → map findings → remediate → re-test), which provides a reasonable workflow. However, there are no explicit validation checkpoints, no error recovery loops, and the re-test step lacks specificity on what constitutes passing. | 2 / 3 |
Progressive Disclosure | The skill references `resources/implementation-playbook.md` for detailed procedures, which is good one-level-deep disclosure. However, the main content is thin enough that it's unclear what value the SKILL.md provides on its own versus just pointing to the playbook, and the reference could be better signaled with more context about what the playbook contains. | 2 / 3 |
Total | 7 / 12 Passed |