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.
The body delivers a clear, well-structured TDD discipline workflow with explicit red-green-refactor checkpoints and an enforcement table, and it stays mostly concise. Its main weakness is actionability: the command blocks are placeholder comments rather than executable commands, and the deprecation notice conflicts with the detailed actionable guidance that follows.
Suggestions
Replace the empty bash comment blocks with a concrete example test command or a clearly parameterized placeholder (e.g. `pytest tests/test_<feature>.py -k <test_name>`).
Reconcile the 'Hard-deprecated. Do not invoke' banner with the substantive guidance below — either move the body into the replacement skill or trim it to a pure redirect.
Trim the 'External Model Consultation' section, which adds tokens without advancing the core test-first discipline.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely lean imperative guidance ('Write ONLY enough code to pass the test', 'Run test - MUST FAIL') with little concept explanation Claude does not already know, matching 'Efficient; minor instances of over-explanation that could be trimmed'. It is not a 5 because the dramatic all-caps enforcement lines and the elaborate 'External Model Consultation' section add tokens that could be trimmed. | 4 / 5 |
Actionability | The Commands section contains empty bash blocks with only comments ('# Run the project's test command - should have ONE new failure') rather than real executable commands, and the Output Format is a template, fitting 'Some concrete guidance but incomplete; pseudocode instead of executable code'. It is not a 4 because no concrete, copy-paste-ready command or code is provided. | 3 / 5 |
Workflow Clarity | The Red-Green-Refactor cycle is a clear sequence with explicit validation checkpoints ('MUST FAIL', 'MUST PASS', 'Must stay green') and an enforcement table for error recovery, matching 'Clear sequence with most checkpoints present; minor validation gaps'. It is not a 5 because the verification commands are abstract placeholders rather than explicit, runnable validation steps. | 4 / 5 |
Progressive Disclosure | The skill has well-organized section headers (Iron Law, Cycle, Enforcement Rules, Commands, Output Format) and no bundle files exist that would require deeper references, matching 'Good structure; most content is appropriately placed'. It is not a 5 because the content runs past 50 lines with a few sections (consultation, output template) that could be trimmed or externalized. | 4 / 5 |
Total | 15 / 20 Passed |