Content
92%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.
Excellent operational content: both invocation modes are fully executable, outcomes and failure modes are tabulated with exit codes and recovery options, and the verified-details-to-references split is textbook progressive disclosure. The sole weakness is mild wordiness in the Overview and Guardrails sections, where the same rationale is stated more than once.
Suggestions
Trim guardrail 2 ('Floor, never round up') to a single clause — 'computed with exact rational arithmetic so contracts * risk_per_contract <= risk_budget holds by construction; no epsilon' — removing the repeated 'by construction' and the redundant restatement of the float-division concern.
Shorten the Overview paragraph explaining why this is a separate skill from position-sizer to one sentence; the detailed 20-1000x risk-error rationale already lives in references/sizing-methodology.md, so a pointer suffices.
Resolve the dangling forward reference in Step 2 ('See the reason glossary below') by either adding a one-line reason glossary under the table or pointing at the no_trade_reason enumeration in the Output Contract.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and skill-specific — it does not re-explain what futures are, and its explanatory passages (the ES/NQ/ZB dollar-per-point contrast, the 110'16 vs 110.16 tick-grid hazard) are verified, non-obvious domain facts rather than padding. It falls short of 5 because a few passages over-explain: guardrail 2 says 'by construction' twice and restates the no-epsilon argument, and the Overview's rationale for being a separate skill is partially repeated in 'When to Use' and the resources notes. | 4 / 5 |
Actionability | Every workflow step is a copy-paste-ready command: full Mode A and Mode B invocations with realistic arguments, the '--list-specs' command, and a WRONG/CORRECT command pair for the bond off-grid case. The Output Contract gives a fully populated ES example with exit-code semantics ('usage error, exit 2', 'exit 0, a report IS written') and concrete recovery advice ('widen the stop, increase risk %, or skip the trade'), covering the common cases. | 5 / 5 |
Workflow Clarity | Steps 1-4 are clearly sequenced (size, read result, check warnings, inspect specs) with explicit checkpoints: a sizing_status table explaining each outcome, a non-blocking warnings section with per-warning meaning, and fail-closed exit-code conventions distinguishing operator-caused from gate-file-caused failures. Error recovery is addressed (what a NO_TRADE from risk_below_one_contract means and the three operator options), matching the anchor-5 pattern; the destructive/batch cap does not apply to a pure calculator. | 5 / 5 |
Progressive Disclosure | SKILL.md stays an overview: the 23-market spec table and methodology/formulas are one level deep in references/futures-contract-specs.md and references/sizing-methodology.md, both real files that match their descriptions, clearly signaled inline ('see references/futures-contract-specs.md for the per-row source URLs') and catalogued in a Resources section that summarizes each file's role. Nothing that belongs in a reference file is inlined, and no nested (2+ level) references exist. | 5 / 5 |
Total | 19 / 20 Passed |