Content
22%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is essentially a table of contents with no substantive content. It lists high-level steps for building backtesting systems but provides zero executable code, no concrete examples, no specific patterns, and no validation checkpoints. All meaningful content is deferred to referenced files that aren't provided, making the SKILL.md itself nearly useless as standalone guidance.
Suggestions
Add at least one concrete, executable code example in the SKILL.md itself — e.g., a minimal event-driven backtest loop skeleton in Python showing the core pattern.
Include explicit validation checkpoints in the workflow, such as verifying no look-ahead bias by checking that signals only use data available at each timestamp, and checking that transaction cost assumptions are realistic before running full backtests.
Replace the abstract instruction bullets with specific, actionable steps — e.g., instead of 'Build point-in-time data pipelines', show a concrete pattern for ensuring point-in-time correctness with a code snippet.
Consolidate the reference paths — the skill references both `resources/implementation-playbook.md` and `./sub-skills/implementation-playbook.md` which appear to be different files or a naming inconsistency that should be clarified.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 'Use this skill when' and 'Do not use this skill when' sections add moderate value but are somewhat verbose for Claude. The instructions themselves are lean bullet points, but the overall structure includes some padding (e.g., the 'Safety' section states obvious things Claude already knows about not giving financial advice). | 2 / 3 |
Actionability | The instructions are entirely abstract and descriptive — 'Define hypothesis, universe, timeframe', 'Build point-in-time data pipelines', 'Implement event-driven simulation' — with zero concrete code, commands, specific examples, or executable guidance. Everything is deferred to a referenced playbook that isn't provided. | 1 / 3 |
Workflow Clarity | While the five bullet points in Instructions loosely imply a sequence, there are no explicit validation checkpoints, no feedback loops, and no concrete steps for a multi-step process that involves complex, error-prone operations like backtesting infrastructure. The workflow is essentially 'do these things' without any verification or error recovery. | 1 / 3 |
Progressive Disclosure | The skill references `resources/implementation-playbook.md` and a sub-skill, which is a reasonable one-level-deep structure. However, no bundle files are provided, so we can't verify the references exist. The SKILL.md itself is too thin — it's essentially just a pointer to other files with almost no standalone value, and the sub-skills section at the bottom references a different path than the resources section, creating confusion. | 2 / 3 |
Total | 6 / 12 Passed |