Content
71%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.
Well-structured instruction skill with a clear sequenced workflow, a concrete output template, and exemplary progressive disclosure via a Load-When reference table pointing at six real one-level-deep files. Its main weakness is token efficiency: it re-teaches concepts Claude already knows (N+1, magic numbers, SQL injection) and closes with a redundant topic list, both of which could be trimmed or moved to references.
Suggestions
Cut or drastically shrink the 'Review Patterns (Quick Reference)' section: N+1 queries, magic numbers, and SQL injection bad/good examples re-teach concepts Claude already knows — a one-line prompt ('check for N+1, injection, magic numbers — see references/common-issues.md for patterns') would preserve the trigger at a fraction of the tokens.
Replace the 'Knowledge Reference' line ('SOLID, DRY, KISS, YAGNI, design patterns, OWASP Top 10...') with either nothing or one actionable instruction (e.g., 'Apply OWASP Top 10 as the security baseline'), since a bare topic list gives Claude no guidance it doesn't already have.
Add an explicit validation checkpoint to the Tests step (e.g., 'If a behavior change has no covering test, flag it as a critical issue') so every workflow phase has a concrete gate rather than only step 1.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient, but the 'Review Patterns (Quick Reference)' section spends three full code blocks illustrating N+1 queries, magic numbers, and SQL injection — concepts Claude already knows — and the 'Knowledge Reference' line ('SOLID, DRY, KISS, YAGNI, design patterns...') is a bare topic list that adds no information. This is more than the minor trimming of a 4, fitting 'some unnecessary explanation or could be tightened'. | 3 / 5 |
Actionability | The guidance is mostly executable: a numbered 5-step workflow with concrete questions ('Are there N+1 queries, hardcoded secrets, or injection risks?'), a 7-section Output Template, and explicit MUST DO / MUST NOT constraints. It falls short of a 5 because some steps defer detail ('Apply the checks in the Reference Guide below') and the Knowledge Reference gives topics without any actionable instruction. | 4 / 5 |
Workflow Clarity | A clear 5-step sequence with an explicit early checkpoint ('Summarize the PR's intent in one sentence before proceeding. If you cannot, ask the author to clarify') and mid-process escalation ('If critical issues are found in step 3, note them immediately'). Only one formal checkpoint exists and the test-review step has no validate-feedback loop, matching 'clear sequence with most checkpoints present; minor validation gaps'. | 4 / 5 |
Progressive Disclosure | The Reference Guide table signals six real, verified reference files with explicit 'Load When' triggers, all one level deep with no nested references, while the body stays a concise overview. This matches the top anchor of a clear overview with well-signaled one-level-deep references. | 5 / 5 |
Total | 16 / 20 Passed |