Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A high-quality, actionable skill body with clear phased workflow, validation checkpoints, and a smart progressive-disclosure decision to externalize deterministic metric counting into a script. The only drag is mild redundancy between the phase sections and the Gotchas list.
Suggestions
De-duplicate the data-threshold rules: state 'Pattern mode needs ≥5 meetings' and 'Outcome correlations need ≥3 per group' once in Phase 2b and reference them from Gotchas instead of restating both.
Tighten the philosophical framing lines (e.g. 'The point is not to roast you. The point is to give you a kind, evidence-based mirror...') since the Gotchas already encode 'coaching, not roasting' as an actionable rule.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient with concrete commands, a results table, and output templates, but carries mild redundancy — the 'Pattern mode needs ≥5 meetings' and 'Outcome correlations need ≥3 per group' thresholds are restated in both Phase 2b and the Gotchas section, and the coaching-philosophy lines ('The point is not to roast you') add tokens beyond what Claude needs. | 2 / 3 |
Actionability | Fully executable guidance throughout — exact bash invocations ('minutes voices --json', the mirror_metrics.py call with the --self flag), a complete output-field table, copy-paste output-format templates, and documented exit codes for error recovery. | 3 / 3 |
Workflow Clarity | Clear phased sequence (Phase 0 → 1 → 2a/2b → 3) with explicit validation checkpoints: exit-code-3 handling, minimum data thresholds, and 'If thresholds aren't met, surface what you can compute and tell the user explicitly' feedback loops. | 3 / 3 |
Progressive Disclosure | Well-organized single-file body with one appropriately extracted bundle script (scripts/mirror_metrics.py, verified present) referenced by a clean one-level-deep path via ${CLAUDE_PLUGIN_ROOT}; the exact-counting logic is correctly pulled out of context rather than done in-context, with no nested references. | 3 / 3 |
Total | 11 / 12 Passed |