Content
75%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, actionable workflow with concrete commands, explicit validation checkpoints, and a real error-recovery feedback loop. It is concise and assumes Claude's intelligence; the only thing keeping it from the top band is that some search/extraction steps delegate to other skills without concrete detail.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence (no explanations of what investors or databases are), using tables and compact step lists; the only padding is the intro line repeating the frontmatter description, keeping it just short of a 5. | 4 / 5 |
Actionability | Provides copy-paste-ready bash commands with full argument flags (init-db.sh, check-investor.sh, record-investor.sh) and concrete field-extraction lists, but the search/extract steps defer to other skills ("使用 smart-search 构造搜索 URL") without showing how, a minor gap. | 4 / 5 |
Workflow Clarity | Clear Step 1–4 sequence with an explicit dedup validation checkpoint (skip if {"exists": true}), a browser retry feedback loop (wait 30s → retry → reopen → stop), a match-criteria checklist, and a final statistics summary; the batch cap is not triggered because validation is present, but per-record write verification is only implicit. | 4 / 5 |
Progressive Disclosure | Well-organized into clear sections (前置条件, 搜索渠道, 执行流程, 错误处理) with clearly signaled one-level-deep cross-skill references and no nested bundle references; at ~107 lines it exceeds the simple-skill (<50 line) exception that would allow a 5, so it sits at 4. | 4 / 5 |
Total | 16 / 20 Passed |