Content
45%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 well-organized but generic restatement of textbook TDD that adds little Claude doesn't already know, with no concrete commands, code, or failure-recovery guidance. The cycle and phase rules give it decent workflow clarity, but token efficiency and actionability are the weak points.
Suggestions
Cut sections that restate known doctrine (Three Laws, AAA pattern, YAGNI, generic anti-patterns) and keep only project-specific deviations or hard rules.
Add concrete, executable guidance: the actual test command(s) and framework for this project, an example failing test, and how to run the suite in each phase.
Add failure-recovery loops to the cycle (e.g. what to do if the RED test passes unexpectedly, or a refactor breaks a green test) and remove the vacuous trailing "When to Use" section that duplicates the description's job.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Nearly the entire body restates TDD doctrine Claude already knows — the three laws of TDD, YAGNI ("You Aren't Gonna Need It"), the AAA pattern, and generic anti-pattern tables ("Test implementation | Test behavior") — none of which is skill-specific knowledge. This matches level 2 ('noticeably verbose; several unnecessary explanations or padded sections'); it does not reach level 3 because the unnecessary sections are the majority of the file, not isolated instances, though the tabular formatting keeps it short of 'heavily padded' level 1. | 2 / 5 |
Actionability | The body only describes principles ("Test must fail first", "One assertion per test", "Commit after each refactor") without any concrete, executable guidance: no test commands, no framework, no code example, no project-specific steps. As an instruction-only skill this is not automatically penalized for lacking code, but the guidance is high-level hints rather than specific steps, matching level 2 ('minimal concrete guidance; high-level hints but missing the specific steps to execute') rather than level 3, which requires some concrete guidance beyond restated principles. | 2 / 5 |
Workflow Clarity | The RED → GREEN → REFACTOR → repeat cycle is clearly sequenced in section 1, and each phase has explicit checkpoint rules ("Test must fail first", "All tests must stay green", "Small incremental changes"), matching level 4 ('clear sequence with most checkpoints present; minor validation gaps'). It falls short of level 5 because there are no feedback/recovery loops (e.g. what to do when the RED test unexpectedly passes or a refactor breaks a test) and no concrete validation commands. | 4 / 5 |
Progressive Disclosure | The body is well-organized into ten numbered sections with clear headers, no nested or buried references, and the single-file structure at ~145 lines is navigable, matching level 4 ('good structure; most content appropriately placed; minor organization gaps'). It does not reach level 5 because the trailing duplicated "When to Use" section is vacuous ("This skill is applicable to execute the workflow or actions described in the overview.") and sections like the AI-augmented multi-agent pattern sit at the same level as core workflow content without clear signposting of what is essential. | 4 / 5 |
Total | 12 / 20 Passed |