Content
86%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 a tight, well-structured instruction skill that assumes Claude's competence and gives a concrete output contract. Adding a worked finding example and a brief verify-before-tag checkpoint would round out actionability and workflow clarity.
Suggestions
Show one worked finding line (e.g., 'stdlib: hand-rolled clamp. Use math.Clamp. [util/math.go]') to make the output format copy-paste concrete.
Add a short checkpoint in the Hunt section to confirm a candidate is genuinely dead/unused before tagging it, reducing false positives in a repo-wide scan.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient with no concept explanations Claude doesn't need; every line earns its place ('Deps the stdlib or platform already ships, single-implementation interfaces, factories with one product'). | 5 / 5 |
Actionability | Concrete output format ('`<tag> <what to cut>. <replacement>. [path]`') and a tagged replacement taxonomy give actionable guidance, though no worked example of a finding is shown. | 4 / 5 |
Workflow Clarity | Clear single-pass sequence (Hunt → ranked Output) with explicit boundaries and a revert command, but no checkpoint verifying a candidate is real before tagging it in a repo-wide scan. | 4 / 5 |
Progressive Disclosure | Under 50 lines with no need for external references and well-organized sections (Tags, Hunt, Output, Boundaries), qualifying for 5 under the simple-skill exception. | 5 / 5 |
Total | 18 / 20 Passed |