Content
75%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, concise, actionable skill body with a clear tool-selection decision path and executable examples. All dimensions land at 4 rather than 5 due to small gaps: no error-recovery guidance, undescribed JSON output shape, and no external reference file for the adapter list.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — comparison tables, short code blocks, a decision tree — and assumes Claude's competence without explaining known concepts, matching the 'efficient; minor instances of over-explanation' anchor. It isn't 5 because the '核心优势' (core advantages) section partially repeats the description's claims in a promotional tone ('比 browser_task 的 "让 LLM 猜" 可靠得多') and could be trimmed. | 4 / 5 |
Actionability | It provides concrete, copy-paste-ready calls (`opencli_run(command="zhihu hot list")`, `npm install -g @jackwener/opencli`, `opencli_doctor()`) plus the command format, matching 'mostly executable guidance with minor gaps'. It isn't 5 because the `args` parameter is shown once without explanation, output JSON structure is never described, and `opencli_doctor` is listed with no guidance on interpreting its result. | 4 / 5 |
Workflow Clarity | The 决策路径 decision tree, prerequisites checklist, and tool-discovery flow (opencli_list → opencli_run → opencli_doctor) give a clear sequence with most checkpoints present, matching the anchor for 'clear sequence with most checkpoints'. It isn't 5 because there is no error-recovery guidance (what to do when a command fails or doctor reports a missing Browser Bridge), though the operations shown are read-heavy rather than destructive so the destructive-cap rule doesn't apply. | 4 / 5 |
Progressive Disclosure | The skill is self-contained with well-organized sections and no nested references (no bundle files exist), matching 'good structure; most content is appropriately placed'. It isn't 5 because at ~65 lines it exceeds the under-50-line simple-skill threshold, and the growing supported-site/adapter list plus error-handling detail would be better placed in a one-level-deep reference file. | 4 / 5 |
Total | 16 / 20 Passed |