Content
77%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.
The body is a highly actionable, well-sequenced analysis procedure with strong validation gates and feedback loops, and almost every line carries domain knowledge Claude would not already have. Its weaknesses are repetition of the same rules across multiple sections and a monolithic structure that would benefit from splitting the Log Map, output template, and DMP procedure into bundled reference files.
Suggestions
Consolidate the partNN split-archive rules, the DMP analysis rules, and the zh_cn.json localization guidance so each is stated fully in one place (Workflow) and only briefly cross-referenced in Reminders and Output Format, cutting significant duplicate tokens.
Move the Log Map and the full Output Format markdown template into bundled reference files (e.g. references/log-map.md and references/output-template.md), keeping SKILL.md as a concise overview with clearly signaled one-level-deep links.
Add a natural crash-related trigger term (e.g. 崩溃/dmp) to the description so the DMP analysis capability is discoverable when users report crashes.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Nearly all content is non-obvious domain knowledge (log-to-component mapping, signal names, repo correspondence) with no basic-concept padding, but the partNN split-archive rules, the DMP rules, and the zh_cn.json localization guidance are each repeated two to three times across Workflow, Output Format, and Reminders, which could be consolidated. It is not 4 because the repetition is noticeable rather than minor; it is not 2 because nothing explains concepts Claude already knows. | 3 / 5 |
Actionability | Guidance is copy-paste executable throughout: 'git clone --depth 1 --branch "v<VERSION>" https://github.com/MaaXYZ/MaaFramework.git .cache/upstream-src/MaaFramework', 'curl -L' to '.cache/issue-logs/issue-<number>/', exact log signals ('Tasker.Task.Starting', 'Node.Recognition.Failed', 'post_task returned task_id'), concrete paths, and a worked blob-link example. Minor vagueness (no named unzip tool for split archives) does not undermine coverage of the common cases. | 5 / 5 |
Workflow Clarity | The 8-step workflow has explicit validation checkpoints and feedback loops: all parts must '存在且非空，才能开始解压', incomplete parts trigger returning to the issue text ('不得解压；应返回 issue 正文和评论继续查找'), the DMP PID is cross-validated against maafw.log '[Px<pid>]', a missing DMP section mandates '返回步骤 5 补做', and maintainer conclusions must be independently re-verified. Not 4: checkpoints are explicit and recovery loops are stated for every fragile operation. | 5 / 5 |
Progressive Disclosure | Section headers are clear and external skill references are well-signaled and one level deep ('.agents/skills/dmp-analysis/SKILL.md', '.agents/skills/tasks/'), but no bundle reference files exist and the ~460-line body is monolithic — the Log Map, the long Output Format template, and the DMP procedure are content that clearly belongs in separate reference files. It is not 4 because substantial separable content is inlined; it is not 2 because navigation via headers is still easy. | 3 / 5 |
Total | 16 / 20 Passed |