Content
25%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 well-organized rule index whose entire value proposition — loading detail from rules/*.md files — is broken, because none of the 17 referenced files exist in the bundle. Combined with a fully redundant resource re-listing and a closing note pointing to non-existent IF-THEN rules, the skill provides navigation to nowhere and no standalone executable guidance.
Suggestions
Ship the actual rules/ directory (all 17 files referenced by the tables) or remove the links — a progressive-disclosure skill whose every reference is dangling is unusable.
Delete the duplicated "📦 可用资源" section; the rule tables already list every file, so it is ~20 lines of pure token waste.
Replace the note about non-existent "IF-THEN 规则" with an explicit lookup procedure, e.g. "Match the user's error message or symptom to the Keyword column, then load the linked rule file and follow its fix."
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The two rule tables are lean, but the entire "📦 可用资源" section (17 bullet lines re-listing every rules/*.md path) is pure duplication of the table links, and the closing note "根据 SKILL.md 中的 IF-THEN 规则判断是否需要加载" references IF-THEN rules that do not exist in the body. That is a substantial padded section, matching "Noticeably verbose; several unnecessary explanations or padded sections". It is not 1 because the core tables carry real, non-redundant information; it is not 3 because roughly a third of the body is duplicated or dangling content rather than minor tightening. | 2 / 5 |
Actionability | The body is a pure index: rule names, keywords, and one-line descriptions like "Catch undefined components in templates" and "Fix Volar 3.0 upgrade issues" with no code, commands, or concrete fix steps anywhere — "Minimal concrete guidance; high-level hints but missing the specific steps to execute". It is not 1 because the table rows do give actionable pointers (rule file, matching keywords, fix intent); it is not 3 because none of the promised detail is actually loadable — the referenced rules/ files do not exist in the bundle, so the guidance never becomes executable. | 2 / 5 |
Workflow Clarity | No process is stated at all: there is no sequence (e.g., "match the user's error to keyword column → load the rule file → apply the fix" is never written), no checkpoints, and the only steering note points to non-existent "IF-THEN 规则". This fits "Rough sequence present but many gaps; steps poorly defined; validation absent" — the table structure implies a lookup pattern but nothing is explicit. It is not 1 because the keyword→rule tables do form a coherent, if implicit, navigation affordance. | 2 / 5 |
Progressive Disclosure | Although references are well signaled and one level deep (rule tables linking rules/*.md), every one of the 17 referenced files is missing from the bundle — there is no rules/ directory, so the entire disclosure structure is dangling. Scoring against the actual bundle structure, broken references plus the redundant skill:// re-listing match "Minimal structure; ... references are buried" in effect: navigation leads nowhere. It is not 3 because the problem is not unclear signaling but total non-existence of the referenced content; it is not 1 because the body itself is sectioned into coherent, link-organized tables rather than a monolithic wall. | 2 / 5 |
Total | 8 / 20 Passed |