Content
50%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.
A well-organized domain reference with genuinely useful signal-construction logic and a concrete output template, but it explains finance concepts Claude already knows, inlines reference material that belongs in bundle files, and — critically — provides no executable path to the flow data the whole skill depends on, leaving its guidance illustrative rather than actionable.
Suggestions
Add an executable flow-data retrieval method (e.g., a specific API endpoint, yfinance-derived volume/NAV approximation, or a documented scraping command) so the analysis pipeline can actually be run, and make etf_flow_signal() self-contained or explicitly mark its inputs as illustrative.
Move the ETF category tables, thematic ETF lists, and data-source table into a references/ file (e.g., references/etf-universe.md) and keep SKILL.md as a lean overview with well-signaled one-level-deep pointers.
Trim the creation/redemption mechanics explanation and sector-sensitivity commentary to the flow-specific interpretation rules Claude cannot infer, and remove hard-coded dates and stale AUM figures or move time-sensitive values to a dated data section.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The flow-signal thresholds, breadth formulas, and output template earn their tokens, but the creation/redemption mechanics section and the sector-sensitivity tables largely restate finance knowledge Claude already has, and stale AUM figures plus a hard-coded yfinance date range ('2025-01-01' to '2026-03-30') are time-sensitive data with no old-patterns/deprecated section — mostly efficient with some unnecessary explanation, not the noticeably padded 2. | 3 / 5 |
Actionability | The breadth calculation and yfinance snippet are concrete, but etf_flow_signal() references undefined daily_flows/aum variables, and the skill's core dependency — actual flow data — is answered only by a table of website names with no API endpoints, URLs, or commands, leaving key execution details missing. Not a 4 because the central data-retrieval step is not executable. | 3 / 5 |
Workflow Clarity | The section ordering (concepts → data access → signals → output format) implies a sequence but no explicit step order exists, and validation checkpoints (e.g., sanity-checking flow magnitudes against AUM, flagging quarter-end rebalancing days) are absent — matching the anchor of sequence present but checkpoints missing or implicit. Not a 2 because the structure and output template make the intended process reasonably legible. | 3 / 5 |
Progressive Disclosure | Headers and tables are internally well organized, but the 267-line body inlines the ETF reference tables, data-source table, and output template that clearly belong in separate reference files, and no bundle files exist at all — fitting the anchor of some structure with content that should be separate kept inline. Not a 2 because the document is genuinely navigable rather than a structureless wall of text. | 3 / 5 |
Total | 12 / 20 Passed |