Content
85%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, governance-oriented skill with clear sequenced gates, validation feedback loops, and clean progressive disclosure to real reference files. The main lever is tightening length/repetition and adding a few more concrete executable specifics.
Suggestions
Reduce repetition of authority-chain and claim-ceiling concepts across the gates; state them once and reference back to keep the body leaner.
Add a concrete, copy-paste-ready example of the target-profile record fields filled in for one common mode (e.g. edit) to raise actionability.
Consider consolidating the repeated '失败返回…' phrasing into a single templated gate format to trim tokens without losing the feedback loops.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense, domain-specific guidance that assumes Claude's competence and avoids basic-concept padding, but is long with some concept repetition (authority, claim ceiling) across gates, sitting below the leanest 5 anchor. | 4 / 5 |
Actionability | As an instruction-only skill it provides concrete, specific guidance — named record fields, gate inputs/outputs, evidence taxonomy, and failure signals — with only minor gaps versus fully copy-paste-ready instruction. | 4 / 5 |
Workflow Clarity | A clearly sequenced process (路由→权威链→基础→gates→合同→证据) where each D10 gate is a checklist with trigger/inputs/actions/evidence/failure-return feedback loops, matching the top anchor. | 5 / 5 |
Progressive Disclosure | The body is an overview with well-signaled, one-level-deep references to real bundle files (references/patterns.md, references/tool-profiles.md) under an explicit 按需读取 section, with content appropriately split. | 5 / 5 |
Total | 18 / 20 Passed |