Content
92%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-structured routing-and-governance document: highly actionable with executable flagged commands, clear sequenced workflows with validation/feedback loops for destructive and batch operations, and clean one-level-deep progressive disclosure into 29 verified reference files. Its only weakness is length and some repetition of the confirmation/preauthorization rules across sections.
Suggestions
Consolidate the preauthorization/confirmation rules into a single authoritative section and have other sections reference it by name, rather than restating the '禁止预授权判定底线' conditions in '高频路径', '应用协作者', and '高影响动作'.
The intent-routing table is very wide and dense; consider splitting the long '用户意图' cells or moving per-intent detail fully into the linked references so the table scans faster.
A few cross-skill links (../lark-shared/SKILL.md, creative-design/creative-design.md, ../lark-drive/SKILL.md) point outside this bundle; a one-line note that these are sibling skills in the plugin would make the navigation expectation explicit.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and almost entirely domain-specific knowledge Claude does not already have (auth/scope handling, app_type routing, resource-vs-source boundaries, destructive-op gates), so most tokens earn their place; however it is long and the preauthorization/confirmation rules recur across '高频路径', '应用协作者', and '高影响动作', which could be consolidated. | 4 / 5 |
Actionability | Provides copy-paste-ready, fully-flagged commands throughout — 'lark-cli apps +member-add --app-id <app_id> --member-type openid --member-id ou_xxx --perm view --dry-run --as user', '+metric-list --app-id <app_id> --metric requests --since <range> --as user', '+get --app-id <meta_token> -q ".data.app.app_id"' — covering the common cases concretely rather than abstractly. | 5 / 5 |
Workflow Clarity | Multi-step flows are explicitly sequenced with validation checkpoints and feedback loops: the numbered app_id-acquisition fallback, the deploy chain 'git commit + push -> +release-create -> +release-get 轮询', and the git error recovery '报错 -> +git-credential-init -> 重试 -> 失败则停止并报告'; destructive/batch operations carry explicit '禁止预授权判定底线' dry-run gates. | 5 / 5 |
Progressive Disclosure | SKILL.md is a routing/governance overview pointing one level deep to per-command references via the '按需读取' column and inline links; the 29 referenced ./references/*.md files all exist, content is appropriately split (overview here, per-command detail in references), and navigation is easy through the intent-routing table. | 5 / 5 |
Total | 19 / 20 Passed |