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.
An exemplary lean skill body: it documents only what Claude cannot infer (the script's interface, endpoint paths, request patterns) and provides three fully executable JSON examples plus a runnable shell one-liner. Structure is tight, the single bundle reference is real, and failure and staleness behavior are covered without any padding.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 38-line body contains only operational facts Claude would not know (base_url, max_items heuristics, stdin field list, truncation caveat) with zero explanation of known concepts. Lean throughout; anchor 5 fits and 4 would require over-explanation to trim, which is absent. | 5 / 5 |
Actionability | The 'echo ... | python scripts/rest_request.py' command plus three complete copy-paste-ready JSON payloads (gene, clinicalAnnotation with params, variant) are fully executable and cover the common lookup cases. Anchor 5 fits; 4 would imply gaps or non-runnable examples that are not present. | 5 / 5 |
Workflow Clarity | This is a simple, single-purpose skill (build request JSON, run script, summarize output) and that single action is unambiguous, with failure behavior documented ('Failure returns ok=false with error.code and error.message') and staleness guidance ('Re-run requests in long conversations'). The simple-skill exception applies and no destructive or batch operations exist, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | The skill is under 50 lines, organized into clear sections, references exactly one bundle file (scripts/rest_request.py, which exists on disk), and explicitly states the package should stay limited to SKILL.md plus that script. Per the simple-skill exception, well-organized sections with no external references warrant anchor 5. | 5 / 5 |
Total | 20 / 20 Passed |