Content
82%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 is a highly actionable, executable reference for the OLS utility scripts with a clearly sequenced two-step resolution workflow and proper one-level reference separation. Weaknesses are modest: redundant guidance across the Recipes/Utility/Workflow sections, no error-recovery path for failed or empty searches, and an unreferenced citation.bib in the bundle.
Suggestions
Consolidate the overlapping 'Quick Recipes', 'Workflow', and per-script documentation into one navigation pattern (e.g., keep recipes + options, fold the Workflow section's restated guidance into it) to remove redundancy.
Add a brief error-recovery note for zero-hit searches (e.g., when --exact returns nothing, retry without --exact or with a synonym) instead of the absolute 'do NOT search across multiple ontologies' rule.
Reference references/citation.bib in the body (e.g., in a Reference section alongside api_reference.md) so every bundle file is discoverable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational — copy-paste commands, option tables, no explanations of concepts Claude already knows. Minor trimming opportunities remain: the "Quick Recipes", "Utility Scripts", and "Workflow" sections restate overlapping guidance, and the "--output is required for all scripts" callout is repeated per-script (and contradicted by per-script "default: stdout" entries). Not level 5 ('every token earns its place') given this redundancy, but well above the noticeably-verbose level 2-3 range. | 4 / 5 |
Actionability | Every script has fully executable, copy-paste-ready commands (e.g., `uv run scripts/search_ols.py --query "diabetes" --rows 5 --output /tmp/ols_search_results.json`), all flags documented with concrete values, and worked multi-step examples use real IDs (DOID:5844, GO:0005634). Covers the common cases comprehensively. | 5 / 5 |
Workflow Clarity | The multi-step entity-resolution pattern is explicitly sequenced ("exactly 2 steps": search with --exact --rows 1, then get relations) and the Workflow section gives a clear 8-step decision sequence. Operations are read-only API queries, so the destructive/batch validation cap does not apply, but there are no checkpoints or feedback loops for failure cases (e.g., zero-hit searches — the "do NOT search across multiple ontologies" rule forbids any fallback), leaving it at 'most checkpoints present; minor validation gaps'. | 4 / 5 |
Progressive Disclosure | Good structure: option docs and recipes are appropriately inline for a script-driven skill, and the API reference is clearly signaled and one level deep ([references/api_reference.md](references/api_reference.md), verified to exist). However, bundle file references/citation.bib is never mentioned anywhere in SKILL.md, a minor organization gap that keeps it below the 'easy navigation, content appropriately split' top anchor. | 4 / 5 |
Total | 17 / 20 Passed |