Content
100%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.
This is a model instruction-only SKILL.md body: a tight five-step cycle with validation checkpoints and feedback loops, rules that resolve real edge cases, a completion checklist, and a single well-signaled one-level-deep reference. Nothing is padded, nothing is missing for the workflow it teaches.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~26-line body contains zero concept explanation and zero padding; every sentence is an instruction ('Write one test for one behavior', 'Add nothing the test does not need'), assuming Claude's competence throughout. | 5 / 5 |
Actionability | Although this is an instruction-only skill with no code, the guidance is fully directive and covers the common cases: the five-step cycle, a bug fix that 'starts with a test that reproduces the bug', the pass-on-red failure mode ('If it passes, it tests nothing new: change the test'), and the hard-test design signal. Per the scoring notes, instruction-only skills are not penalized for lacking code when the guidance is this actionable. | 5 / 5 |
Workflow Clarity | The Red/Green/Refactor sequence is explicit with validation checkpoints at every step ('Watch it fail', 'Watch it pass... no errors or warnings'), an error-recovery feedback loop ('must fail because the behavior is missing, not because of a typo'), and a closing 'Before you say it is done' checklist — exactly the anchor-5 pattern. | 5 / 5 |
Progressive Disclosure | The skill is under 50 lines with well-organized sections (cycle, rules, done-checklist), and its single reference is real (references/writing-good-tests.md exists), one level deep, and clearly signaled with a load condition ('When writing or changing a test, read...'). | 5 / 5 |
Total | 20 / 20 Passed |