Content
71%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 skill body with genuinely runnable commands, a clear step sequence, and exemplary one-level-deep reference signaling to real bundle files. Weaknesses are redundancy (duplicated capability list and dependency declaration), narrated rather than executable examples, vague error-recovery guidance, and leftover malformed lines at the end of the file.
Suggestions
Conciseness: delete the stray lines and dangling code fence after 示例 3 (SKILL.md lines 103-106), and state the Pillow dependency once (frontmatter or 前置准备, not both).
Conciseness: replace the 任务目标 capability bullet list, which restates the frontmatter description, with a single-sentence purpose statement.
Workflow clarity: turn the API-failure branch into an explicit recovery loop (check the error message → verify API key and image format → re-run the script) instead of the current vague '提供错误信息和可能的解决方案'.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient, but the 任务目标 capability list duplicates the frontmatter description, the Pillow dependency is declared twice (frontmatter and 前置准备), the examples are padded with dialogue-style comments, and stray trailing lines ("# 1. 调用脚本分析图片..." plus a dangling fence) remain after 示例 3. Not 4: these are more than minor trimmable instances; not 2: the body is not padded with concepts Claude already knows. | 3 / 5 |
Actionability | Concrete, runnable commands are given ("python scripts/analyze_design.py --image ./uploads/landing-page.png") and both scripts' inputs/outputs are documented with flags (--input/--output). Not 5: 示例 2 and 示例 3 are narrated comment outlines rather than copy-paste-ready commands, and no sample output is shown. | 4 / 5 |
Workflow Clarity | A clear three-step sequence (接收图片 → 调用分析脚本 → 展示结果) with optional branches including API failure and a result-usage path. Not 5: the failure branch only says to provide error info and possible solutions without an explicit validate/retry checkpoint loop; not 3: the sequence and branch conditions are explicit, not implicit. | 4 / 5 |
Progressive Disclosure | The body is an overview that points one level deep to real bundle files — scripts/analyze_design.py, scripts/convert_to_roadshow_style.py, references/api-spec.md, references/roadshow-export-guide.md (all verified to exist) — each annotated with its purpose, parameters, or a "何时读取" (when to read) trigger, plus a "仅在需要时读取参考" note. This matches the 5 anchor exactly. | 5 / 5 |
Total | 16 / 20 Passed |