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.
A well-crafted instruction-only skill body: dense, assumption-respecting prose with specific review moves and concrete reporting requirements ('State the structural fix, not "add a comment"'). The only notable gap is the absence of a worked example illustrating what a good finding looks like.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~35-line body is lean and assumes Claude's competence — no basic-concept explanations, no padding; every line instructs (e.g., 'a helper, schema or convention the codebase already has and this one rebuilds'). It matches the anchor: every token earns its place. | 5 / 5 |
Actionability | For an instruction-only skill the guidance is specific and executable: 'Read each new or changed name against what the code behind it does', 'read the signature and its description as a consumer who cannot see the implementation', 'name the precedent with its file path'. It falls short of 5 only because no worked example of a finding is given to cover common cases. | 4 / 5 |
Workflow Clarity | The Method section sequences the review (read the AGENTS.md/CLAUDE.md chain, then names vs. behavior, then public boundaries, then contract enforcement) and includes a checkpoint ('Before suggesting reuse, verify that the existing code is the appropriate precedent'). This is a read-only skill so the destructive-operation validation cap does not apply; it misses 5 for lacking an explicit feedback/recovery loop. | 4 / 5 |
Progressive Disclosure | Under 50 lines, no external references needed, and no bundle files exist; the body is cleanly organized into Scope, Method, Threshold, and Reporting sections. Per the simple-skill guideline, this earns a 5 with well-organized sections alone. | 5 / 5 |
Total | 18 / 20 Passed |