Content
57%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 skill is a clean, brief overview with excellent progressive disclosure into references/rule.md, but its instructional core is thin: known-concept padding fills the intro, and the Check/Fix guidance lacks inline tools or executable detail to act on directly. Adding a validation checkpoint (e.g., verifying new tests run and fail appropriately in CI) would strengthen the workflow.
Suggestions
Replace the introductory explanation of unit-vs-integration testing with task-relevant guidance (e.g., how to pick which workflows qualify as 'key', or how to size a test boundary), keeping the pedagogical comparison out of the body.
Inline one short, framework-agnostic runnable example or a concrete command (e.g., how to run the integration suite and assert it gates CI), rather than deferring all executable detail to the reference.
Add a validation step to the Check/Fix sequence — e.g., 'confirm the new integration tests fail against the un-fixed code and pass after, and that CI runs them on every PR' — to close the feedback loop the Code Review section asks reviewers to look for.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is short and its sections are terse, but the intro paragraph ("Unit tests verify individual functions in isolation, but they can't catch the bugs that occur when those functions interact") and the first two Quick Reference bullets re-explain the unit-vs-integration distinction Claude already knows. Trimming that pedagogical padding would move it to lean. | 3 / 5 |
Actionability | Some concrete anchors exist ("Use a real database in tests (test DB or in-memory) rather than mocking everything", "Test happy paths and the most critical error paths", and named integration-point categories in Check), but Fix and Explain stay high-level with no tools, commands, or code inline — the executable detail is entirely deferred to references/rule.md. | 3 / 5 |
Workflow Clarity | The Check → Fix → Explain → Code Review sections give a coherent rough sequence for the review workflow, but there are no validation checkpoints or feedback loops (e.g., how to confirm the written tests run in CI or that failures block regressions), even though the Code Review section itself is about enforcement gaps. | 3 / 5 |
Progressive Disclosure | The body is a compact, well-sectioned overview that appropriately pushes implementation detail to a single clearly-signaled, one-level-deep reference ("For full implementation details, code examples, and framework-specific guidance, see references/rule.md"), which exists and matches that description. Navigation is easy with no buried or nested references. | 5 / 5 |
Total | 14 / 20 Passed |