csp-skill-optimizerlisted

| Signal | Meaning | Action | |--------|---------|--------| | **Missing trigger** | Skill should have activated but didn't | Add trigger words to description | | **Missing rule** | Skill activated but wrong behavior | Add specific rule/red flag | | **Missing context** | Skill didn't account for project-specific pattern | Add project-aware section | | **Rationalization loophole** | Agent found workaround for existing rule | Close loophole explicitly | | **Stale content** | Skill references outdated patterns | Update with current patterns | | **Token bloat** | Skill too large, slow to load | Compress, cross-reference, split | **Analysis process:** 1. For each feedback signal, identify which skill (if any) should have handled it 2. If no skill covers this → **new skill needed** or **extend existing skill** 3. If a skill covers but failed → **gap in that skill's rules** 4. If a skill succeeded but user wanted different → **user preference to capture** ### Pillar 3: Optimization Engine (Act) Apply optimizations based on gap analysis: **Optimization types:** ``` Type 1: Trigger Refinement - Update description field with new trigger words - Add synonym coverage for common user phrasings Type 2: Rule Addition - Add new rule based on user correction - Add red flag for the rationalization pattern observed Type 3: Context Injection - Capture project-specific conventions - Add patterns observed in successful implementations Type 4: Loophole Closure - When agent viol