Content
75%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.
The body is highly actionable with complete executable commands for every mode and a clear sequenced workflow backed by real, one-level-deep bundle files. Its main weakness is token efficiency: general trading principles and dated regulatory details are inlined in SKILL.md instead of being trimmed or moved to the methodology reference.
Suggestions
Trim the Key Principles section to skill-specific rules and move general trading wisdom (loss asymmetry, half-Kelly rationale, portfolio heat) into references/sizing_methodologies.md, which already covers risk management principles.
Move the dated FINRA intraday-margin details out of the main body — either into the methodology reference or a dedicated dated/regulatory section — so time-sensitive specifics don't penalize the core instructions.
Condense the 28-line JSON report example to the fields needed to interpret output, and add an explicit validation step to the workflow (e.g., check the binding constraint and re-run with tightened limits when a constraint overrides the risk-based size).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient, but the Key Principles section recites trading knowledge Claude already has ('A 50% loss requires a 100% gain to recover', 'Survival first'), and the FINRA rule embeds time-sensitive dates ('effective 2026-06-04... through 2027-10-20') outside any deprecated/old-patterns section, which the guidelines explicitly penalize. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready CLI commands cover all five invocation modes (fixed fractional, fractional shares, ATR, Kelly budget, Kelly shares) plus a constraints example, with every flag and value specified — this matches the top anchor of covering the common cases completely. | 5 / 5 |
Workflow Clarity | The six-step workflow is clearly sequenced (gather parameters → execute → load reference → scenarios → constraints → report) with an input constraint ('--stop must be positive and below --entry'), but there is no explicit validation or error-recovery checkpoint on the script output. This is a non-destructive calculation skill, so the destructive/batch cap of 3 does not apply. | 4 / 5 |
Progressive Disclosure | The SKILL.md is a well-organized overview with one-level-deep, clearly signaled references ('references/sizing_methodologies.md', 'scripts/position_sizer.py'), both of which exist in the bundle. However, methodology content that belongs in the reference file (Half Kelly, portfolio heat, the 1% rule) is inlined in Key Principles, leaving minor organization gaps. | 4 / 5 |
Total | 16 / 20 Passed |