Content
80%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A dense, actionable skill body with executable CLI guidance and clear rule tables. The main gaps are the absence of an explicit error-recovery feedback loop in the workflow and the failure to point readers to the existing reference files.
Suggestions
Add an explicit feedback loop to the workflow, e.g. after step 2 (validate data quality): if sessions are skipped or volume is unreliable, re-run with a longer lookback or flag the run as unreliable before proceeding.
Link the reference files from the body — e.g. under Core Rules add 'See [references/ibd_distribution_methodology.md](references/ibd_distribution_methodology.md) for full rule derivation' and under TQQQ Exposure Policy point to references/tqqq_exposure_policy.md.
Surface the scripts/ entry point and module roles (e.g. risk_classifier.py, exposure_policy.py) so a reader knows where each workflow step is implemented without listing the directory.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean, table-driven body that assumes Claude's competence — it states rules and thresholds without explaining what a distribution day or EMA is, and every section earns its place. | 3 / 3 |
Actionability | Provides a copy-paste-ready CLI invocation plus concrete thresholds (0.2%, 25 sessions, 5%), named API functions, and exact output paths rather than abstract direction. | 3 / 3 |
Workflow Clarity | The 9-step workflow is clearly sequenced and includes a data-quality validation step, but there is no explicit validate→fix→retry feedback loop for the batch data operation, leaving checkpoints implicit. | 2 / 3 |
Progressive Disclosure | The body is well-sectioned and detailed methodology/policy are split into references/ files, but those reference files are never linked or signaled from the body, so navigation to them is missing. | 2 / 3 |
Total | 10 / 12 Passed |