Content
65%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A concrete, example-rich quant skill with strong actionability and clear section structure, held back by re-explanation of basic concepts, absence of validation feedback loops, and a monolithic layout with no reference files. Tightening basic-concept exposition and splitting deep-dive material into references would raise the weaker dimensions.
Suggestions
Remove or compress basic-Greek definitions and the 'option price = intrinsic + time value' primer, which Claude already knows, to improve conciseness.
Add explicit validation checkpoints for risky actions (e.g. 'Before opening a calendar spread, confirm near-month IV ≤ far-month IV; if inverted, do not enter') to strengthen workflow clarity.
Move detailed worked examples, market-making basics, and SABR calibration into one-level-deep reference files (e.g. EXAMPLES.md, MARKET_MAKING.md) linked from the overview to improve progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense and value-adding (SABR parameter ranges, Zakamouline hedge triggers, China A-share specifics), but it re-explains basic Greeks Claude already knows (Delta/Vega/Theta/Gamma meanings, 'option price = intrinsic + time value') and could be tightened, matching the 'mostly efficient with some unnecessary explanation' anchor. | 2 / 3 |
Actionability | Provides concrete, specific guidance: numeric thresholds (stop loss at 50% of net debit, max loss = 2x premium, Delta limit ±500), decision formulas (Gamma × S² × σ² × Δt > 2 × transaction_cost), and fully worked 50ETF/300ETF examples with prices and IVs — concrete and copy-paste-ready for an instruction skill. | 3 / 3 |
Workflow Clarity | Each strategy follows a clear Principle → Entry conditions → Example → Risk-control sequence and the Output Format gives a structured deliverable, but there are no explicit validation/feedback-loop checkpoints; risk-control items are reactive stop-losses rather than validate-then-proceed steps, matching the 'sequence present but checkpoints missing or implicit' anchor. | 2 / 3 |
Progressive Disclosure | Sections are well organized (Overview, Core Concepts, Analysis Framework, Output Format, Notes), but the skill is a single ~188-line monolithic file with no bundle references, exceeding the 50-line simple-skill threshold, so content that could be split (worked examples, market-making, SABR calibration) is inline. | 2 / 3 |
Total | 9 / 12 Passed |