Content
57%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.
A well-organized, code-dense pytest reference that is highly actionable in most sections, but it functions as a monolithic ~490-line catalog in SKILL.md with no reference files, includes several sections of basic pytest knowledge Claude already has, and offers no sequenced workflow or validation checkpoints.
Suggestions
Split reference-catalog content into bundle files (e.g. references/fixtures.md, references/mocking.md, references/hypothesis.md) and keep SKILL.md as a concise overview with clearly signaled one-level-deep links.
Cut sections that restate knowledge Claude already has — test discovery conventions, basic assertion syntax, and the elementary 'pytest -v / -s' run flags — keeping only non-obvious patterns and project-specific conventions.
Add a short sequenced workflow with checkpoints, e.g. write test → run 'pytest --lf' on failures → check 'pytest --cov --cov-report=term-missing' → act on missing lines, so users get an ordered process instead of a topic catalog.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean pattern code rather than padded prose, but it spends entire sections on pytest fundamentals Claude already knows — 'pytest automatically discovers tests following these conventions', a 'Basic Assertions' block showing 'assert value == expected', and basic run flags like 'pytest -v' and 'pytest -s'. This is more than a minor instance of over-explanation (ruling out 4) yet the bulk is legitimately useful pattern reference rather than filler (ruling out 2). | 3 / 5 |
Actionability | Nearly every section provides concrete, executable code or commands — complete fixture examples with setup/teardown, parametrize tables, coverage commands ('pytest --cov=src --cov-report=term-missing'), and marker-based run commands. Minor gaps keep it below 5: examples reference undefined symbols (User, UserRepository, validate_email), and 'mock_email_service.send.assert_called_once()' is subtly wrong after patching the class (it needs '.return_value.send'). | 4 / 5 |
Workflow Clarity | The content is a topical reference with no sequenced workflow — there is no 'write test → run → check coverage → improve' pipeline, and no validation or feedback loop for the workflows it implies (e.g., what to do when '--cov-fail-under=80' fails). This matches anchor 3 (steps/structure present but checkpoints missing); the skill is too long (>50 lines) for the simple-skill exception to apply. | 3 / 5 |
Progressive Disclosure | No bundle files exist, so all ~490 lines live inline in SKILL.md. Section headers provide real structure, but reference-catalog content that clearly belongs in separate files — the full run-command catalog, hypothesis property-based testing, advanced mocking — is inlined, matching anchor 3 ('content that should be separate is inline') rather than anchor 4's 'most content appropriately placed'. | 3 / 5 |
Total | 13 / 20 Passed |