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.
A strong operational skill body: fully executable commands and schemas, a clearly sequenced workflow with deterministic validation gates and error-recovery loops, and correct use of one-level-deep references. The only weakness is minor redundancy and rhetorical padding that could be tightened without losing meaning.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body never explains concepts Claude already knows and mostly earns its tokens with rules, commands, and tables. Minor over-explanation remains: a few rules stated twice (the gate appears in Non-Negotiable #5, Step 6, and the Convergence Contract; the noise budget repeats in the severity table) and rhetorical flourishes like 'Training data is a rumor; the changelog is a source' and 'arguing with it is arguing with a regex' could be trimmed. This is efficient with minor trimmable instances, not the lean every-token-earns-its-place level. | 4 / 5 |
Actionability | Guidance is fully executable: exact gh CLI commands ('gh pr view --json number,title,body,author,url...', 'gh api user --jq .login'), concrete script invocations ('python3 scripts/review_gate.py findings.json'), a complete findings.json example with schema fields explained, and worked examples covering routine review, blockers, unevidenced claims, and round-2 re-reviews. Nothing is pseudocode. | 5 / 5 |
Workflow Clarity | Steps 0-7 are clearly sequenced with explicit validation checkpoints and feedback loops: Step 4 verification kills un-evidenced candidates, Step 6 mandates 'Fix every reported violation and re-run until exit code 0', the deterministic ladder runs before any judgment, and a Troubleshooting section maps errors to solutions (422, gh auth, no PR, gate loops). The outward-facing post step is gated behind deterministic validation. | 5 / 5 |
Progressive Disclosure | Structure follows progressive disclosure well: both referenced files exist and are one level deep, clearly signaled at exactly the point of use ('Read `references/review-standards.md` now', 'Read `references/comment-voice.md` now'), and the three scripts are referenced by path and present in scripts/. Detailed standards and comment voice are correctly externalized while the body keeps the overview, workflow, and contract. | 5 / 5 |
Total | 19 / 20 Passed |