Content
67%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-structured, highly actionable API skill whose code examples are executable and match the bundled script exactly. Its main weaknesses are length and repetition — duplicated technical specs, generic Python error-handling tutorials, and a Quick Start path that doesn't match the real bundle layout.
Suggestions
Consolidate the rate limit, max page size, and base URL facts into Technical Specifications only, and trim the generic try/except and '.get()' tutorials in Best Practices to one-line pointers — Claude already knows this Python.
Fix the Quick Start path inconsistency: 'cd scientific-databases/clinicaltrials-database/scripts' should be 'cd scripts' relative to the skill root, matching 'python3 scripts/query_clinicaltrials.py' used later.
Move several of the ten capability walkthroughs (e.g., sponsor search, combined query strategies) into references/api_reference.md, keeping SKILL.md as a lean overview with the strongest 3-4 patterns.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient, API-specific content, but with notable slack: the ~50 req/min rate limit is stated three times (Overview, Best Practices, Technical Specifications), max page size 1000 and base URL are each repeated, and the Best Practices section teaches generic Python Claude already knows (boilerplate try/except requests error handling, '.get()' safe navigation). Not a 2 because the bulk is genuinely novel, domain-specific API guidance rather than padded concept explanation. Not a 4 because the duplication and generic tutorials are more than minor instances. | 3 / 5 |
Actionability | Code examples are concrete and executable: real API parameters ('query.cond', 'filter.overallStatus'), real response paths, and helper functions that all exist in scripts/query_clinicaltrials.py (verified: search_studies, get_study_details, search_with_all_results, extract_study_summary, with matching signatures including format='csv'). Not a 5 because of the Quick Start path bug — 'cd scientific-databases/clinicaltrials-database/scripts' does not match the actual bundle layout (script lives at scripts/query_clinicaltrials.py, invoked as 'python3 scripts/query_clinicaltrials.py' elsewhere in the same file), and the 'from scripts.query_clinicaltrials import ...' examples silently assume running from the skill root. | 4 / 5 |
Workflow Clarity | Clear task-oriented sequencing: Quick Start → capability-specific patterns → pagination/bulk handling → export, with validation guidance present (rate-limit 429 backoff retry loop, raise_for_status error handling, pageToken break condition, missing-data checks). Not a 5 because checkpoints are advisory rather than woven into the workflows themselves — e.g., no step verifying retrieved count against totalCount or confirming response status before parsing in the main examples. Not a 3 because the sequences are coherent and error-recovery guidance is explicitly provided. | 4 / 5 |
Progressive Disclosure | Bundle structure is solid: both referenced files exist (references/api_reference.md, scripts/query_clinicaltrials.py), references are one level deep, and the Resources section clearly signals what each contains and when to load it ('Load this reference when working with unfamiliar API features or troubleshooting issues'). Not a 5 because the SKILL.md itself runs ~500 lines with ten full capability walkthroughs and Best Practices detail that partly duplicate the reference file and could be split out, leaving the main file less of a lean overview than the anchor describes. | 4 / 5 |
Total | 15 / 20 Passed |