Content
81%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 body is well-structured and actionable with a clear single-level reference to a real rule.md file. Its main weakness is mild over-explanation of concepts Claude already knows in the 'Explain' section and the absence of explicit validation checkpoints in the test workflow.
Suggestions
Trim the 'Explain' section: drop the explanation of the accessibility tree and reading vs forms/application mode — Claude already knows these; keep only the screen-reader-specific quirks that are non-obvious.
Add an explicit validation step in the 'Check' workflow, e.g. 'After each fix, re-run the failing screen reader command and confirm the correct name/role/state is announced before moving on.'
Include at least one concrete code snippet in 'Fix' (e.g. a roving tabindex pattern or an aria-live region) so the guidance is copy-paste ready rather than prose.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely lean and well-organized into Quick Reference / Check / Fix / Explain sections, but the 'Explain' section describes concepts Claude already knows (the accessibility tree, reading vs forms/application mode, virtual cursor), which is mild over-explanation that could be trimmed; this is the 4-anchor 'minor instances of over-explanation'. | 4 / 5 |
Actionability | It gives concrete, executable guidance — specific screen reader + browser combos, key commands (H key, D/R keys, Insert+F7), and six numbered fixes with attribute names (aria-live='polite', roving tabindex, aria-describedby); minor gaps are that there is no copy-paste code snippet and the fixes are prose rather than examples, landing at 4. | 4 / 5 |
Workflow Clarity | The 'Check' section sequences the test workflow clearly (choose 2 combos, navigate by keyboard, verify name/role/state, check dynamic content, manage focus) and 'Fix' enumerates issues with remedies, but there is no explicit validation checkpoint confirming the fix resolved the announcement, which is the 4-anchor 'clear sequence with most checkpoints present; minor validation gaps'. | 4 / 5 |
Progressive Disclosure | SKILL.md is a concise overview with well-signaled one-level-deep reference to references/rule.md (which exists as a real file), section headers organize the content, and detailed material is offloaded to that single reference, matching the 5-anchor clear-overview pattern. | 5 / 5 |
Total | 17 / 20 Passed |