Content
96%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.
An excellent instruction-only skill body: every tool call is concrete with exact parameters, the decision workflow is unambiguous with explicit failure handling, and there is no token waste on concepts Claude already knows. The only minor gap is a slightly dangling build-script reference and a body a touch over the compact-skill line count.
Suggestions
Trim or clarify the generated-file header comment: the referenced scripts/generate-cherry-assistant-knowledge and pnpm build target do not exist in this bundle, so either drop the comment or note it applies only to the source repo.
Consider moving the five "回答时遵守以下规则" interpretations into a short reference file if the skill grows, keeping SKILL.md under the ~50-line compact threshold.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence: it never explains what MCP or a manifest is, and every sentence adds a constraint or instruction Claude could not infer (e.g., "routes.all 还可能包含内部页、参数路由或兼容跳转，不能无条件推荐"). No padded or tutorial-style sections exist, matching the lean-and-efficient anchor. | 5 / 5 |
Actionability | Fully executable guidance: exact tool calls with parameters are given for each question type (e.g., mcp__assistant__product_info({ source: "manifest", section: "routes" })), plus a concrete index-first fallback and a numbered rule list for interpreting results. This is copy-paste-ready for an instruction-only skill. | 5 / 5 |
Workflow Clarity | The decision procedure is clearly sequenced with explicit checkpoints: question→section mapping, index call when the section is unknown, navigate in the same turn before manual steps, and failure handling ("工具未调用或失败时，不得声称已经生成入口"; fallback to official docs when the manifest lacks the capability). The only deliberate restriction on section: "all" prevents context blowup, showing error-avoidance rather than missing validation. | 5 / 5 |
Progressive Disclosure | Well-organized single-level sections with no nested references, and content is appropriately inlined for a compact instruction skill. It falls just short of the top anchor because the body slightly exceeds the ~50-line simple-skill threshold and the header comment cites a generator script (scripts/generate-cherry-assistant-knowledge) that is not present in the bundle, leaving a dangling reference. | 4 / 5 |
Total | 19 / 20 Passed |