Content
52%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 has a sensible skeleton (when to use, instructions, safety) and a clear sequenced workflow, but it is undermined by a truncated opening code block, an orphaned "Role" convention no step uses, refactor examples that contradict the skill's stated scope, and a Resources pointer to a file that does not exist.
Suggestions
Create resources/implementation-playbook.md (or remove the reference) so the Resources section points to a real file.
Repair the opening Django code block (missing fence/heading and imports) and remove the refactor-phase examples or move them to the playbook, since the skill explicitly excludes refactoring work.
Delete the "Steps marked **Role**..." line — no step is marked Role — and add a concrete test-run command plus a retry step when tests still fail.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The prose is tight with no over-explanation of TDD concepts, but the body carries refactor-phase code variants ("Refactor: Class-based view", "Refactor: Full layering") that contradict the skill's own "Do not use this skill when... refactoring" scope and waste tokens. Not 4 because this off-scope content is unnecessary material that should be trimmed. | 3 / 5 |
Actionability | Real Django and Express code plus a 4-step instruction list provide concrete guidance, but the opening Django block is truncated (no opening fence, missing imports), Express examples reference undefined `users`/`userService`, and no actual test-run command is given. Not 4 because these gaps go beyond minor: the first example is malformed and the instructions stay high-level ("identify the smallest fix"). | 3 / 5 |
Workflow Clarity | The Instructions section gives a clear sequence (review failing tests, implement minimal change, run tests, record debt) with an explicit validation checkpoint ("Run tests after each change to confirm progress"). Not 5 because there is no error-recovery loop for when tests still fail and no concrete test command; not 3 because a checkpoint is explicitly present, not implicit. | 4 / 5 |
Progressive Disclosure | The only external reference, "resources/implementation-playbook.md", does not exist in the bundle, so navigation fails entirely, and the pattern catalogs that the playbook was meant to hold are inlined in SKILL.md. Not 3 because a broken reference is a worse navigation failure than a present-but-unclear one, and the structure has defects (truncated opening block, orphaned "### Express Patterns" heading, "Role" convention with no matching steps). | 2 / 5 |
Total | 12 / 20 Passed |