---

name: performance-optimizer description: "Identify and resolve performance bottlenecks through profiling, measurement, and targeted optimization." autoInvoke: false priority: medium triggers:

• "performance optimization"
• "slow"
• "bottleneck"

---

Skill: Performance Optimizer

Category: Dev Expert Version: 1.0.0 Used By: All development agents

---

Overview

Identify and resolve performance bottlenecks through profiling, measurement, and targeted optimization.

---

1. Performance Optimization Rules

1. Measure first, optimize second
2. Optimize the bottleneck, not everything
3. 80/20 rule: 80% of slowness is in 20% of code
4. Premature optimization is the root of all evil

---

2. Profiling Strategy

Layer	Tools	Metrics
Frontend	Lighthouse, DevTools	FCP, LCP, TTI, CLS
Backend	APM, profilers	Response time, throughput
Database	EXPLAIN, slow query log	Query time, index usage
Memory	Heap snapshots	Allocation, leaks

When to Profile

• Before optimization (baseline)
• After each change (verify improvement)
• Production (real user data)

---

3. Frontend Optimization

Core Web Vitals Targets

Metric	Good	Description
LCP	< 2.5s	Largest Contentful Paint
FID	< 100ms	First Input Delay
CLS	< 0.1	Cumulative Layout Shift

Quick Wins

// Lazy load images
<img loading="lazy" src="image.jpg" />

// Code splitting
const Component = lazy(() => import('./Component'))

// Debounce handlers
const handleSearch = debounce((q) => search(q), 300)

// Memoize expensive computations
const result = useMemo(() => expensiveCalc(data), [data])

Checklist

• Images optimized (WebP, lazy loading)
• Code split by route
• CSS/JS minified and compressed
• Caching headers set
• Fonts preloaded

---

4. Backend Optimization

Common Bottlenecks

Issue	Solution
N+1 queries	Eager loading, batching
Missing indexes	Add appropriate indexes
Unbounded queries	Pagination, limits
Sync blocking	Async/parallel processing
No caching	Cache hot data

Quick Wins

// Batch database calls
const users = await User.findByIds(ids)  // 1 query, not N

// Add indexes
CREATE INDEX idx_users_email ON users(email)

// Cache expensive queries
const data = await cache.getOrSet('key', () => db.query(), 3600)

// Parallel async
const [users, orders] = await Promise.all([getUsers(), getOrders()])

---

5. Database Optimization

Query Analysis

-- Always EXPLAIN slow queries
EXPLAIN ANALYZE SELECT * FROM orders WHERE user_id = 123;

-- Look for: Seq Scan (bad), Index Scan (good)

Index Strategy

Query Pattern	Index Type
Exact match	B-tree (default)
Range queries	B-tree
Full-text search	GIN/GiST
JSON fields	GIN

Checklist

• Indexes on WHERE/JOIN columns
• No SELECT * (specify columns)
• Pagination on large tables
• Connection pooling enabled
• Query result caching

---

6. Caching Strategy

Cache Level	TTL	Use Case
Browser	Hours-Days	Static assets
CDN	Minutes-Hours	API responses
Application	Seconds-Minutes	Computed data
Database	Varies	Query cache

Cache Invalidation

// Time-based
cache.set(key, value, { ttl: 3600 })

// Event-based
onUserUpdate(user => cache.delete(`user:${user.id}`))

// Version-based
cache.set(`data:v${version}`, value)

---

7. Memory Optimization

Common Leaks

Cause	Fix
Event listeners not removed	Cleanup in useEffect/destroy
Closures holding references	Null out references
Growing arrays/maps	Use WeakMap, clear periodically
Timers not cleared	clearInterval/clearTimeout

Detection

// Node.js heap snapshot
const v8 = require('v8')
v8.writeHeapSnapshot()

// Browser DevTools
// Memory tab → Take heap snapshot → Compare

---

8. Optimization Checklist

Before:

• Baseline metrics captured
• Bottleneck identified via profiling
• Target improvement defined

After:

• Metrics improved
• No regressions introduced
• Tests still pass

---

Best Practices

Do's

• Measure before and after
• Optimize biggest bottleneck first
• Use appropriate data structures
• Cache strategically
• Profile in production-like env

Don'ts

• Optimize without measuring
• Micro-optimize everything
• Cache without invalidation strategy
• Ignore memory leaks
• Sacrifice readability for speed

---

Version: 1.0.0 | Last Updated: 2025-11-28