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.
The body is an exemplary lean reference: executable quick-start code, dense tables of non-obvious API facts, and honest limitations with per-case workarounds, plus a clearly quarantined retirement notice. The only structural gap is that all reference material lives inline in SKILL.md rather than being split into a bundle file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section carries non-obvious, project-specific facts (frequency codes, prefix-to-exchange rules, the `vol` → `volume` rename, 800-row pagination limits, 前复权 default, the httpx pin workaround) with no padding explaining known concepts, and time-sensitive retirement details are quarantined in a status/deprecated block — matching the lean anchor 5. | 5 / 5 |
Actionability | Quick Start gives copy-paste-ready executable code (`Quotes.factory(market="std")`, `get_k_data(code=..., start_date=..., end_date=...)`, `bars(symbol=..., frequency=1, offset=800)`) covering the common daily and intraday cases, and the Key Methods table adds signatures and return schemas — matching anchor 5. | 5 / 5 |
Workflow Clarity | This is a single-task data-source skill and the single action is unambiguous: preferred API per use case, the over-fetch-then-clip intraday strategy, and explicit failure paths with alternatives ("Loader logs a warning and skips BJ symbols — use akshare or tushare") — the simple-skill exception applies and no destructive/batch validation is needed. | 5 / 5 |
Progressive Disclosure | The body has good structure with clear section headers and labeled external links ("Mootdx 文档", "通达信协议参考"), but there are no bundle files and all ~97 lines — including a Key Methods API table and frequency-code table that could be a separate reference file — are inlined, so it is a minor organization gap short of anchor 5. | 4 / 5 |
Total | 19 / 20 Passed |