Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a strong, highly actionable multi-step research workflow with executable tooling and robust validation/audit feedback loops, held back mainly by verbosity and meta-overhead and by reliance on out-of-bundle references instead of bundled detail files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is mostly specialized investment-research methodology Claude would not inherently produce in this form (four-masters framing, rigor gates), but it carries meta-overhead (the Codex adapter note), repeated data-source listings, and verbose bias tables that could be tightened, so it is not fully lean. | 2 / 3 |
Actionability | It provides concrete, parameterized executable commands — e.g. `python3 tools/financial_rigor.py verify-market-cap --price {股价} --shares {总股本} --reported {报告市值}`, plus cross-validate, verify-valuation, three-scenario, and report_audit subcommands — that are copy-paste ready once placeholders are filled. | 3 / 3 |
Workflow Clarity | A clearly sequenced 前置步骤 + seven-step process is reinforced with explicit validation checkpoints and feedback loops ("如果工具报告 ❌ 偏差过大，必须排查原因后才能继续分析" and the 数据抽检 准出/打回 retry loop), matching the validate→fix→retry pattern. | 3 / 3 |
Progressive Disclosure | Sections are well-organized and a detail spec is signaled one level deep ("> 数据源规范：参见 skills/financial-data.md"), but no bundle files exist and referenced paths (skills/financial-data.md, tools/) point outside the skill, so the structure is monolithic rather than a clean in-bundle progressive disclosure. | 2 / 3 |
Total | 10 / 12 Passed |