autoclaw-performance

name: autoclaw-performance description: Audit PANaCEa performance. Identify slow queries, large bundles, inefficient rendering, and optimization opportunities. Use when app feels slow or before production changes. mode: performance

Performance Mode — Optimization Audit

Purpose

Find and fix performance issues. Always measure before optimizing.

When to Use

• App feels slow (page load, interactions, test runs)
• Build times increasing
• Bundle size growing
• Before production deployment
• When adding new dependencies

Audit Targets

Build & Bundle

npm run build 2>&1 | grep -i "size\|chunk\|warning"
du -sh dist/

• Bundle under reasonable size?
• Code splitting working?
• Dead code eliminated?
• Dependencies not duplicated?

Database

• Missing indexes on frequent WHERE/JOIN columns?
• N+1 queries in API endpoints?
• Unbounded queries (missing LIMIT)?
• Analytics queries using appropriate aggregations?

Frontend

• Large component re-renders (React DevTools Profiler)?
• Unnecessary state updates?
• Images optimized?
• Lazy loading for heavy components?
• Debouncing on user input?

API

• Response payloads trimmed to needed fields?
• Caching headers set where appropriate?
• Rate limiting functional?
• AI calls batched/cached?

Output Format

## Performance Audit

### Findings
1. 🔴 {critical bottleneck} — {location} — {impact} — {fix}
2. 🟡 {improvement opportunity} — {location} — {expected gain}

### Bundle (if relevant)
- Before: {size}
- After: {size}
- Savings: {delta}

### Queries (if relevant)  
- Slowest: {query} — {time} — {fix}

### Recommendations
1. {highest-impact fix first}
2. {next}

Coordination

• Triggered by: Orchestrator (pre-deploy), Builder (suspected perf issue)
• Hands off to: Architect (for redesign needed), Builder (to apply optimizations)
• Stored in: .autoclaw/performance-notes.md
• Measure first: Never optimize without data — profile before touching code

Common Pitfalls

• Premature optimization: Only optimize measured bottlenecks
• Micro-optimizations: Focus on N+1 queries, large bundles, blocking ops first
• Build analysis: npm run build output shows chunk sizes — check for regressions