Content
57%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 content is a well-structured, actionable Stock Connect flow-analysis reference with strong executable code and a clear output template, but it is somewhat verbose, lacks explicit workflow checkpoints, and is a monolithic single file with no progressive disclosure into reference files.
Suggestions
Tighten or move the 'Why Northbound/Southbound matters' explanatory prose into a reference file, keeping only the signal-relevant thresholds inline to improve conciseness.
Add an explicit numbered analysis workflow (pull data → compute signals → score composite → render output) with a validation checkpoint (e.g. sanity-check quota utilization and cumulative-flow signs) to lift workflow clarity.
Split the data-access API reference and the composite-signal scoring logic into separate reference files (e.g. references/data_access.md, references/composite_signal.md) and link to them from SKILL.md to enable one-level-deep progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and information-dense (signal thresholds, sector tables, executable code), but the 'Why Northbound matters' / 'Why Southbound matters' prose blocks and the 7-bullet Notes commentary include conceptual explanation that could be tightened; it is not 4 because several explanatory passages go beyond what is strictly actionable, and not 2 because the padding is moderate rather than severe. | 3 / 5 |
Actionability | Provides executable Python for signal functions, Tushare moneyflow_hsgt/hsgt_top10 calls, yfinance downloads, and a concrete output-format template — mostly copy-paste ready — but minor gaps (the composite connect_score is a commented dict rather than a function, and northbound_trend returns None on some branches) keep it just below fully-executable 5. | 4 / 5 |
Workflow Clarity | The logical flow (concepts → signals → data access → output format) and the structured output template give an implicit sequence, but there are no explicit numbered steps or validation checkpoints; this is an analytical rather than destructive/batch skill so the destructive cap does not apply, and it is not 4 because checkpoints remain implicit rather than present. | 3 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent) and the ~240-line skill is a single monolithic document; it has good section headers and structure, but detailed code, data-access reference, and the output template are all inlined rather than split into one-level-deep references, fitting 'some structure but content that should be separate is inline' rather than 4. | 3 / 5 |
Total | 13 / 20 Passed |