Content
88%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, highly actionable skill body: exact I/O contracts, concrete input patterns, and a runnable example with clear error and truncation handling appropriate to a simple read-only lookup. Its one real defect is that the References section misdescribes the bundle by omitting the variant_resolution.py dependency, plus slight rule duplication across sections.
Suggestions
Update the References section to acknowledge scripts/variant_resolution.py as part of the bundle (tpmi_phewas.py imports resolve_query_variant from it), so the stated file inventory matches reality.
Merge the duplicated max_results and save_raw guidance from 'Operating rules' and 'Execution behavior' into a single location to remove redundancy and tighten token usage.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — no concept explanations, just rules, I/O contracts, and one example. Minor trimming is possible: max_results and save_raw guidance each appear redundantly across "Operating rules" and "Execution behavior" (e.g., "start with max_results=10 and only increase it" vs "Increase max_results gradually"), keeping it just below the level 5 'every token earns its place' anchor. | 4 / 5 |
Actionability | Fully executable guidance: exact script path, stdin JSON contract, four concrete input patterns with real values ("{\"grch38\":\"6:160540105-T-C\",\"max_results\":10}"), enumerated output field names, failure shape (ok=false with error.code/message), and a copy-paste bash example covering the common case. Matches the level 5 anchor. | 5 / 5 |
Workflow Clarity | This is a simple single-task skill (~40 lines) whose single action is unambiguous: resolve the one accepted input to a canonical GRCh38 query, run scripts/tpmi_phewas.py, report the canonical variant, count, and truncation status. Error recovery and truncation feedback paths (ok=false/error, truncated -> raise max_results; save_raw for full payloads) are explicit, and the operation is read-only so the destructive/batch validation cap does not apply. Fits the simple-skill level 5 exception. | 5 / 5 |
Progressive Disclosure | The body is under 50 lines, well sectioned, and correctly references scripts/tpmi_phewas.py which exists in the bundle. However, the References section's claim to "keep the import package limited to this file and scripts/tpmi_phewas.py" is inaccurate against the actual bundle: scripts/variant_resolution.py is present and is imported by tpmi_phewas.py yet is never mentioned, a minor navigation/organization gap that keeps it below level 5. | 4 / 5 |
Total | 18 / 20 Passed |