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.
A tight, fully operational SKILL.md: it documents the script interface completely with copy-paste examples, output shapes, and failure handling, while adding zero filler. The simple-skill exceptions in the rubric apply cleanly, so workflow clarity and progressive disclosure earn top marks despite no multi-step workflow or external reference files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is purely operational ("Use `scripts/clinicaltrials_client.py` for all ClinicalTrials.gov v2 calls", "Study searches are better with `max_items=10` and `max_pages=1`") with zero background explanation of what ClinicalTrials.gov is; every line teaches non-obvious rules or I/O shape, matching the lean/efficient anchor. | 5 / 5 |
Actionability | An executable bash invocation plus three copy-paste-ready JSON payloads (studies with filters, metadata, field_values) covering the common cases, a full action list, all optional fields, and explicit failure semantics ("Failure returns `ok=false` with `error.code` and `error.message`"); not 4 because nothing concrete is missing for the documented actions. | 5 / 5 |
Workflow Clarity | The single action (pipe one JSON object to the script) is unambiguous, and the rubric's simple-skill exception applies; error feedback is defined via the failure channel, and re-run/truncation rules guard long conversations. Not a destructive or batch operation, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | Under 50 lines with well-organized sections (Operating rules, Execution behavior, Input, Output, Execution, References); the body correctly states no additional references are required, and the one referenced bundle file, `scripts/clinicaltrials_client.py`, exists, so the simple-skill exception for a 5 holds. | 5 / 5 |
Total | 20 / 20 Passed |