/profile - Performance Profiling

Analyze code for performance characteristics and bottlenecks.

Usage

/profile [target] [--type cpu|memory|io|all]

Arguments:

• target - File, function, or module to profile
• --type - Profiling focus (default: all)

Examples

/profile src/services/dataProcessor.ts
/profile handleRequest --type cpu
/profile src/api/ --type memory

What It Analyzes

CPU Performance

• Algorithm complexity (Big O)
• Hot paths and loops
• Recursive call patterns
• Synchronous blocking operations

Memory Usage

• Memory allocation patterns
• Potential memory leaks
• Object retention issues
• Large data structure handling

I/O Operations

• Database query patterns
• File system operations
• Network call efficiency
• Caching opportunities

Async Patterns

• Promise handling
• Callback patterns
• Event loop blocking
• Concurrency issues

Output Format

## Performance Profile: dataProcessor.ts

### Summary
| Metric | Assessment |
|--------|------------|
| CPU Efficiency | Medium |
| Memory Usage | High Risk |
| I/O Pattern | Suboptimal |
| Overall | Needs Optimization |

### Hotspots Identified

#### 1. processLargeDataset() - Line 45
**Issue:** O(n²) complexity due to nested loops
**Impact:** High - scales poorly with data size
**Current:**
```typescript
items.forEach(item => {
  const match = allItems.find(i => i.id === item.refId);
});

Optimized:

const itemMap = new Map(allItems.map(i => [i.id, i]));
items.forEach(item => {
  const match = itemMap.get(item.refId);
});

Improvement: O(n²) → O(n)

2. Memory Leak Risk - Line 78

Issue: Event listener not cleaned up Impact: Medium - memory grows over time

Recommendations

1. Implement caching for repeated database queries
2. Use streaming for large file processing
3. Add connection pooling for database
4. Consider pagination for large result sets

Benchmarks

Operation	Current	Target
Process 1K items	450ms	<100ms
Memory peak	256MB	<64MB

## Skill Used

`performance-profiling`

## When to Use Agent Instead

Use the **Software Architect** agent when:
- System-wide performance review needed
- Architecture changes may be required
- Capacity planning involved