Content
78%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 highly actionable, well-structured skill body: executable commands for every mode, explicit failure semantics, and exemplary progressive disclosure into two real, well-scoped reference files. The main costs are cross-section repetition of the same rules (mode-B authority, NO_TRADE handling) and a promised reason glossary that is only partially delivered, which drag conciseness and workflow clarity slightly down.
Suggestions
State the mode-B precedence rule once (e.g., in Step 1) and have the Overview and Guardrail 1 reference it, instead of restating 'no --direction/--stop alongside --gate-json' in three places.
Deliver the promised reason glossary: replace Step 2's 'See the reason glossary below' with a compact table of all no_trade_reason values (risk_below_one_contract, gate_not_ready, gate_symbol_mismatch, and any others beyond the output contract's '...'), each with a one-line meaning.
Trim Guardrails 2-3 phrasing -- 'exact by construction, no epsilon' followed by 'no epsilon nudge, no float-representation edge case' says the same thing twice, and the NO_TRADE remediation in Step 2 overlaps Guardrail 2's zero-contracts discussion.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with genuinely non-obvious domain content (32nds notation, tick-grid guards, fail-closed exit-code conventions) and never explains concepts Claude already knows, but rules are repeated across sections: the mode-B authority rule appears in the Overview ("never accepts an explicit --direction/--stop alongside --gate-json"), Step 1 ("--direction/--stop are rejected alongside --gate-json (usage error, exit 2)"), and Guardrail 1, while NO_TRADE handling is covered in both Step 2 and Guardrail 2, and Guardrail 2 says "exact by construction, no epsilon" then repeats "no epsilon nudge, no float-representation edge case". This is 'some unnecessary explanation or could be tightened' rather than the minor trimming of anchor 4. | 3 / 5 |
Actionability | Copy-paste-ready bash commands for both mode A and mode B, the --list-specs invocation, a WRONG-vs-CORRECT worked pair with exact prices and the exit-code consequence, and a complete output-contract YAML with field-level semantics. Fully executable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | A clearly sequenced 4-step workflow with per-step validation (sizing_status table, warning semantics, exit-code conventions) and error-recovery guidance (NO_TRADE -> micro contract, reassess risk pct, or skip). Falls short of 5 because Step 2's "See the reason glossary below" only glosses risk_below_one_contract while gate_not_ready/gate_symbol_mismatch are explained in scattered sections, and the output contract truncates the reason list with "..." -- a minor but real validation-detail gap. | 4 / 5 |
Progressive Disclosure | Overview-first SKILL.md with bulk detail correctly externalized: the 23-market spec table with per-row source URLs lives in references/futures-contract-specs.md (139 lines) and formulas/rationale in references/sizing-methodology.md (100 lines), both verified to exist. The Resources section signals every bundle file (including both scripts) with a one-line description, references are one level deep, and no detail that belongs in a reference is inlined. | 5 / 5 |
Total | 17 / 20 Passed |