Content
67%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-structured, highly actionable body with a crisp step pipeline, exact UI specs, and genuinely useful reference files. The main weakness is redundancy: Step 3 restates standard Black-Scholes and payoff math that both Claude and references/strategies.md already cover, and the inline code template duplicates references/bs_code.md.
Suggestions
Cut Step 3 down to a pointer: keep only the non-obvious calendar-spread note (theoretical value cannot be an expiry function) and defer the standard BS formulas and expiry payoff identities to references/strategies.md and references/bs_code.md, which already contain them.
Replace the stubbed JS code template with a one-line pointer to references/bs_code.md plus the required function signatures, since the reference file already provides the copy-paste implementation.
Add a quick validation checkpoint after Step 1 (e.g., confirm strikes are ordered K1 < K2 < K3 and the multiplier matches the underlying before computing), and make browser verification of the standalone HTML unconditional when a browser is available.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The extraction-defaults table, slider/chart specs, and host-probe command earn their tokens, but Step 3 re-derives standard knowledge Claude already has — the Black-Scholes put formula, put-call parity, and expiry payoff identities — which is duplicated again in references/strategies.md. This is 'mostly efficient but includes some unnecessary explanation that could be tightened' rather than the noticeably-verbose score-2 anchor, since the padding is confined to one section and the rest is dense and skill-specific. | 3 / 5 |
Actionability | Mostly executable guidance: a runnable yfinance probe with exception handling, exact slider ranges and defaults, precise chart color/axis specs, and a JS code skeleton. The inline template is stubbed pseudocode ('function normCDF(x) { /* Horner approximation */ }') rather than copy-paste ready, though the flexibility is justified and the executable implementation lives in references/bs_code.md — matching the minor-gaps anchor instead of the fully-executable score-5 anchor. | 4 / 5 |
Workflow Clarity | A clear five-step sequence (extract → identify → compute → render → respond) with real fallback branches: spot resolution cascades (shown price → live quote → middle strike, 'and tell the user you did'), the probe handles failure, and the standalone-HTML path says to 'verify it in a browser when available'. It falls short of the score-5 anchor because checkpoints are hedged and there is no validation of extracted data itself (e.g., strike ordering or multiplier sanity against the screenshot), which is the main real-world failure mode. | 4 / 5 |
Progressive Disclosure | Good structure with well-signaled one-level-deep references — both references/strategies.md and references/bs_code.md exist and deliver exactly what their descriptions promise, with clear guidance on when to read them ('if you're unsure about payoff formula edge cases'). Not a 5 because Step 3's inline formula blocks overlap with strategies.md's per-strategy formulas, so the overview/detail split isn't clean — some content sits in both places. | 4 / 5 |
Total | 15 / 20 Passed |