Content
67%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-structured procedural skill with a genuinely useful project-context prelude, a clear anti-pattern section, and a well-sequenced workflow with checklists. It is held back by Philosophy-section content that restates knowledge Claude already has and by sidecar references that cannot be confirmed against the bundle.
Suggestions
Cut or compress the Philosophy section's explanation of good-vs-bad tests (behavior through public interfaces, implementation coupling) to a one-line principle and delegate the examples entirely to tests.md/mocking.md, which already exist for that purpose.
Add a short recovery branch to the Incremental Loop for when a test will not go green (e.g., expected-failure diagnosis, when to delete and rewrite the test), closing the workflow's only validation gap.
Ensure the five sidecar files referenced in the body are actually bundled with the skill so the one-level-deep reference structure is verifiable in the distributed artifact.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The project-context prelude and horizontal-slices anti-pattern are dense and earn their tokens, but the Philosophy section re-explains fundamentals Claude already knows ("Tests should verify behavior through public interfaces, not implementation details", "Good tests are integration-style... describe _what_ the system does"), which could be tightened. | 3 / 5 |
Actionability | Concrete, actionable instruction-only guidance: the WRONG/RIGHT horizontal-vs-vertical slice diagram, exact questions to ask the user ("What should the public interface look like?"), and copy-ready per-cycle checklists; minor abstraction remains in refactor steps like "Apply SOLID principles where natural". | 4 / 5 |
Workflow Clarity | Planning, Tracer Bullet, Incremental Loop, and Refactor are clearly sequenced with explicit checkpoints ("Never refactor while RED", "Run tests after each refactor step", "suite green before and after"), but there is no recovery guidance for when a test will not go green, leaving a minor validation gap. | 4 / 5 |
Progressive Disclosure | Sidecar references ([tests.md], [mocking.md], [deep-modules.md], [interface-design.md], [refactoring.md]) are well-signaled, inline at point of relevance, and one level deep, and the prelude points to project files rather than inlining them; however, none of the referenced sidecar files are present in the bundle being reviewed, so the split cannot be verified. | 4 / 5 |
Total | 15 / 20 Passed |