Content
62%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured TDD methodology skill with a clear, validated six-step workflow. It is held back by conceptual prose Claude already knows, the absence of worked examples, and a reference file that largely duplicates the body rather than adding depth.
Suggestions
Trim TDD philosophy Claude already knows (e.g. the 'design tool as well as a verification signal' passage) so every remaining token earns its place.
Add one or two worked examples — a short sample test list and a sample failing test with its minimal passing implementation — to make the guidance concrete and copy-paste ready.
Differentiate the reference file from the body so it adds a distinct layer of depth (e.g. move extended examples, edge-case patterns, or anti-patterns into references/054-design-tdd.md) instead of restating the same six-step workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly disciplined and structured with MUST/MUST NOT constraints, but includes TDD philosophy Claude already knows (e.g. "The failing test is a design tool as well as a verification signal. It lets the usage appear before the implementation hardens around an accidental shape.") that could be trimmed, so it is mostly efficient but not fully lean. | 2 / 3 |
Actionability | The six steps with explicit MUST/MUST NOT constraints and a defined report format are concrete and specific, but the skill provides no worked examples (e.g. a sample test list or a sample failing test), stopping short of the copy-paste-ready, example-anchored guidance of a 3. | 2 / 3 |
Workflow Clarity | A clearly sequenced six-step cycle with explicit validation checkpoints (confirm the test fails for the expected reason, run focused verification for green, keep tests green while refactoring), a feedback loop ("If a refactor changes behavior, return to the failing signal and restore correctness"), and a closing report checklist. | 3 / 3 |
Progressive Disclosure | The body is well-sectioned and points to a single one-level-deep reference file that exists, but the SKILL.md body itself is long and detailed while the reference file substantially overlaps the same workflow, so the overview-to-detail split is not clean and the reference does not add a distinct layer. | 2 / 3 |
Total | 9 / 12 Passed |