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 solid, executable skill body whose main weakness is redundancy: 'When to Use', 'Key Features', and parts of 'Implementation Details' repeat the same capability list three times. The core workflow is clearly sequenced and detail is appropriately pushed to the bundled reference and script files, but the SMILES structure-search capability lacks any concrete example.
Suggestions
Merge 'When to Use' and 'Key Features' into a single section — they currently restate the same five capabilities in near-identical wording, and trim 'Implementation Details' bullets that re-explain what the example code already shows.
Add a concrete, runnable SMILES/structure-search example (or state explicitly that the client does not support it) instead of the vague 'where supported by the API/client' phrasing.
Show a one-line invocation example for the bundled scripts/query_chembl.py CLI (e.g., 'python scripts/query_chembl.py molecule --term aspirin') so users can run it directly, and move the bundle-file pointers into a clearly labeled 'References' section rather than tucking them under 'Dependencies'.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is mostly efficient — it does not explain concepts Claude already knows and the example code is tight — but there is real redundancy: the 'When to Use' and 'Key Features' sections restate the same five capabilities in near-identical wording, and several 'Implementation Details' bullets (filtering model, pagination, bioactivity fields) re-explain what the example code already demonstrates. This fits the level-3 anchor ('mostly efficient but includes some unnecessary explanation or could be tightened'); it is not level 2 because there is no conceptual padding or educational fluff, and not level 4 because the duplicated capability lists are a clear trim target. | 3 / 5 |
Actionability | The guidance is mostly executable: a complete, runnable Python example covering molecule search, MW filtering, target lookup, bioactivity, and mechanism retrieval with concrete field names and operators ('pref_name__icontains', 'molecule_properties__mw_freebase__lte=500'), plus an install command. It falls short of level 5 because the SMILES/structure-search capability is described only vaguely ('where supported by the API/client', 'as documented in references/api_reference.md') with no concrete example, and the bundled scripts/query_chembl.py CLI is mentioned but never shown how to invoke. | 4 / 5 |
Workflow Clarity | The example code lays out a clear numbered sequence (1–5: name search, property filter, target lookup, activity retrieval, mechanism retrieval) with one explicit checkpoint ('raise SystemExit("No molecules found for query.")') and a note to handle missing keys, matching the level-4 anchor 'clear sequence with most checkpoints present; minor validation gaps'. It is not level 5 because there are no error-recovery guidance (e.g., handling empty activity/mechanism results or API errors), though it is not capped at 3 since these are read-only queries, not destructive or batch operations. | 4 / 5 |
Progressive Disclosure | The body stays at overview level and correctly pushes detail to real, verified bundle files — references/api_reference.md (which does contain the filter syntax and resource list) and scripts/query_chembl.py — with pointers from the Dependencies section and Implementation Details ('consult references/api_reference.md for the authoritative list'). This matches the level-4 anchor: good structure, references mostly clear, minor organization gaps — the file pointers sit under 'Dependencies' with hedged wording ('optional, if present in this repository') rather than a dedicated, well-signaled references section. It is not level 5 for that reason, but well above level 3 since nothing that belongs in a separate file is inlined and references are one level deep. | 4 / 5 |
Total | 15 / 20 Passed |