performance-audit

name	Performance Audit
description	Use when analyzing performance bottlenecks or optimizing critical code paths. Finds N+1 queries, blocking I/O, memory leaks, and algorithmic complexity. Read-only analysis with optional benchmarking via Bash.
when_to_use	"why is this slow?", "profile this", "find the bottleneck", "N+1 query", "optimize this path"
context	fork
agent	Explore
allowed-tools	Read, Grep, Glob, Bash
disallowed-tools	Edit, Write, NotebookEdit
disable-model-invocation	true

Performance Audit Skill

Analyze codebase for performance issues and optimization opportunities.

Analysis Steps

1. N+1 Query Detection

   • Search for loops containing database calls
   • Identify missing eager loading patterns
   • Check for batch vs single-record operations

2. Blocking Operations

   • Find synchronous I/O in async contexts
   • Identify missing concurrency patterns
   • Check for sequential operations that could parallelize

3. Memory Leak Patterns

   • Look for unbounded caches or collections
   • Identify missing cleanup in event listeners
   • Check for circular references preventing GC

4. Caching Analysis

   • Review existing cache implementations
   • Identify cache-worthy operations (expensive, repeated)
   • Check cache invalidation strategies

5. Algorithm Complexity

   • Review nested loops for O(n²) patterns
   • Check for unnecessary repeated computations
   • Identify opportunities for memoization

Benchmark Commands

# Node.js profiling
node --prof app.js && node --prof-process isolate-*.log

# Python profiling
python -m cProfile -s cumulative script.py

# Go benchmarks
go test -bench=. -benchmem ./...

# Rust benchmarks
cargo bench

Output Format

Report findings with impact assessment:

• Critical: >100ms latency impact
• High: 10-100ms latency impact
• Medium: Measurable but < 10ms
• Low: Micro-optimization opportunity