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.
The body is concise and well-structured for a simple single-purpose review skill, with a clear sequenced process and a usable rating threshold. Its main weakness is actionability: it tells Claude to "analyse" and "output" without concrete methods or examples for finding the listed issue types.
Suggestions
Add a concrete method or prompt pattern for each issue type in "What It Looks For" (e.g., how to detect duplication or dead code) rather than only listing categories.
Give at least one worked example of a rated finding (input snippet → finding → impact score → recommendation) so the 1-10 rating scale is calibrated.
Specify concrete commands for scoping (e.g., `git diff main --stat`) beyond the table examples to make the first process step executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and section-based with no padding or explanation of concepts Claude already knows; every section (usage, scope table, process, examples, related) earns its place. | 5 / 5 |
Actionability | The Process steps are high-level hints ("Analyse scope for issues", "Output findings with recommendations") with only the 1-10 rating and <4 threshold being concrete; no method, commands, or examples for actually locating issues are given. | 2 / 5 |
Workflow Clarity | The four-step process (analyse, rate, output, skip-below-4) is clearly sequenced and the rating threshold acts as a quality gate; it is a read-only review so the destructive/batch validation cap does not apply, though concrete commands are absent. | 4 / 5 |
Progressive Disclosure | At under 50 lines with no bundle files present and no need for external references, the well-organized section structure satisfies the simple-skill exception for a top score. | 5 / 5 |
Total | 16 / 20 Passed |