Content
92%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 lean, script-driven workflow skill: executable commands for every phase, explicit output validation with error-recovery paths for rate limits and network failure, and sound boundary guidance (example engine names are not a whitelist; pushed_at only proves code updates). The only blemish is minor redundancy between the overview and the 注意事项 section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Efficient overall — bullet-style operational facts (SSL fallback, rate-limit partial-status handling, no-git policy) that Claude cannot infer, with no concept-explaining padding. Not 5 because of minor repetition: '脚本本身不调用大模型' appears both in the overview (line 19) and again in 注意事项 (line 90). | 4 / 5 |
Actionability | Fully executable guidance: copy-paste bash commands with exact flags and {TEMP_DIR} placeholders for all three phases, a comment showing the --language variant, an explicit phrase→--since mapping ('周榜'→weekly, '日榜'→daily, '月榜'→monthly), and stated stdout contract (script prints final note path). | 5 / 5 |
Workflow Clarity | Clear phased sequence (config → parse period → fetch/score → discovery → note → report) with explicit validation and feedback loops: '确认 {TEMP_DIR}/github_trending.json 是非空 JSON 数组。为空则看 stderr 诊断', partial-status preservation on API failure ('不要把空结果写成“没有新增项目”'), and network/SSL fallback handling. | 5 / 5 |
Progressive Disclosure | A single well-sectioned SKILL.md serving as the overview; implementation detail is delegated to the three scripts, shared conventions live one level deep and are clearly signaled ('[Agent 运行约定](../_shared/agent-runtime.md)'). Nothing that belongs in a separate reference file is inlined; navigation is trivial at this size. | 5 / 5 |
Total | 19 / 20 Passed |