Content
92%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.
Strong body: an efficient, fully actionable audit workflow with concrete commands, an explicit report format, and correct use of a single one-level-deep reference file for the detailed rules. The only flaw is minor redundancy — the critical/minor severity definitions are stated both in the report template and in the '## Severity' section.
Suggestions
State the critical/minor severity definitions once (in the '## Severity' section) and have the report template's Summary line just reference them (e.g., 'Violations: N (X critical, Y minor — see Severity), removing the duplicated definitions from the template block.
Alternatively, delete the standalone '## Severity' section and keep the definitions only inside the report template, since that is where Claude needs them when writing the summary.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — concrete git command, terse rule-category list, no concept explanations. However, the severity definitions appear twice in near-identical form: in the report template ('Critical = backdrop-filter/backdrop-blur present anywhere, translucent overlay background, hardcoded colors' / 'Minor = missing DesignSystem helper, non-standard animation duration') and again in the '## Severity' section. That minor duplication could be trimmed, matching the score-4 anchor ('efficient; minor instances... that could be trimmed') rather than 5 ('every token earns its place'). | 4 / 5 |
Actionability | Fully executable guidance: a copy-paste git command ('git diff --name-only HEAD~3 -- src/Client/**/*.fs ...'), exact file paths (references/design-rules.md, src/Client/DesignSystem.fs), an enumerated 8-category checklist, and a complete fill-in report template with an example table row. Matches the score-5 anchor: concrete commands plus a specific output format covering the common case. | 5 / 5 |
Workflow Clarity | Six clearly sequenced steps (identify targets → load rules → read DesignSystem.fs → check files → report → offer fix), with an output checkpoint (the structured report with counts) and a gate on the only risky action ('ask if the user wants violations auto-fixed'). The skill is a read-only audit, so the destructive/batch validation cap does not apply, and no sequence gaps remain. Not 4: every step is unambiguous and the risky follow-up is explicitly gated. | 5 / 5 |
Progressive Disclosure | The SKILL.md is a ~50-line overview that defers the full rule set to a single, clearly signaled, one-level-deep reference ('Read references/design-rules.md for the full rule set'), which exists as a real 126-line file; no other bundle content is inlined and no nested references exist. Matches the score-5 anchor for clear overview with well-signaled one-level-deep references. | 5 / 5 |
Total | 19 / 20 Passed |