Content
81%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-constructed skill body: lean, syntax-dense, with a clearly sequenced workflow, confirmation checkpoints, and pre-output self-checks. The two real weaknesses are the undocumented bundled script (discussed but never given a path or invocation) and mild redundancy between the scope-restating sections and the frontmatter description.
Suggestions
Add a short "Script usage" section (or extend the References section) that discloses scripts/main.py and its subcommands with an invocation example, e.g. `python scripts/main.py validate "(aspirin[Title/Abstract]) AND (stroke[MeSH Terms])"` — the Step 5 warning and the LAST_5_YEARS/LAST_10_YEARS instructions currently reference script behavior with no way to run it.
Merge the "When to Use" and "Input Validation" sections into one brief scope statement; both largely restate the frontmatter description and the out-of-scope list, so consolidation would trim tokens without losing information.
Include one worked cross-database example in Step 5 (e.g., the same concept group rendered in Embase Emtree syntax with /exp and .ti,ab.) so the adaptation rules are demonstrated end-to-end rather than only described.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient — no padding explaining what PubMed or Boolean logic is, and syntax is delivered in dense tables and copy-paste blocks. However, the "When to Use" section and the "Input Validation" section substantially restate the frontmatter description (e.g., "This skill accepts any research question, clinical question, PICO framework, or topic"), which could be merged. Minor trimming keeps this at 4 rather than 5. | 4 / 5 |
Actionability | Mostly executable: exact field-tag syntax ("Diabetes Mellitus/drug therapy"[MeSH Terms]), filter table with literal syntax, pre-built clinical query filters as copy-paste blocks, and concrete adaptation rules (.ti,ab., TS=, /exp). The gap: the body references "the validate subcommand" and "LAST_5_YEARS / LAST_10_YEARS filter expressions" but never states the script's path (scripts/main.py) or an invocation command, so a reader cannot execute the validation it is told to run. This is a concrete missing detail, fitting 4 rather than 5. | 4 / 5 |
Workflow Clarity | A six-step workflow with explicit validation checkpoints and feedback loops: ask a focused clarifying question when input is unclear (Step 1), mandatory flagging of unverified MeSH mappings (Step 2), an intermediate mapping check-in that requires user confirmation before proceeding (Step 2b), a manual bracket-verification step after running validate, and a pre-output Quality Checklist plus Hard Rules. This matches the 5 anchor (explicit validation steps, feedback loops, checklists for complex processes). | 5 / 5 |
Progressive Disclosure | The body is a proper overview and both referenced files exist, are clearly signaled ("→ Detailed MeSH hierarchy guidance: references/mesh-structure.md"), and are one level deep — the two references were verified to exist with relevant content. However, the bundle's 26KB scripts/main.py (with pico, mesh, and validate subcommands) is never mentioned or linked anywhere in SKILL.md, leaving a major bundle asset undiscoverable. That organization gap holds this at 4 rather than 5. | 4 / 5 |
Total | 17 / 20 Passed |