patternslisted

# Pattern Effectiveness Dashboard The user wants to see the pattern learning dashboard showing what the system has learned and how effective those patterns are. ## Protocol 1. **Parse arguments**: `$ARGUMENTS` - If `team` is provided: filter to team patterns only (`.orchestray/team-patterns/`). - If a pattern name is provided: show single pattern detail view. - If empty: show the full dashboard (all sections). 2. **Load patterns via MCP:** Call `mcp__orchestray__pattern_find` with: - `task_summary: "all patterns"` (a neutral query that is broad enough not to bias the scoring hard; the skill wants everything) - `max_results: 50` (the server-side maximum) - `min_confidence: 0.0` (no filter) The returned `matches` array contains objects with `slug`, `uri`, `category`, `confidence` (numeric 0-1), `decayed_confidence` (time-weighted, numeric 0-1), `age_days` (integer), `times_applied`, `one_line`, and `match_reasons`. The server already normalizes confidence to numeric and merges Section 22 / Section 30 field aliases — no client-side normalization needed. `decayed_confidence` applies exponential decay: `confidence × 0.5^(age_days / half_life)` where `half_life` defaults to 90 days (configurable via `pattern_decay.default_half_life_days`). `age_days` is measured from `last_applied` (if set) or the pattern file's mtime. Note: `pattern_find` reads `.orchestray/patterns/` only. To include `.orchestray/team-patterns/*.md`, glo