akashmeshram/analyze-performance

Analyze Performance

Identify bottlenecks using the performance-analyzer agent—no benchmarks required.

What It Finds

1. Algorithmic Complexity - O(n²) loops, inefficient searches
2. Blocking I/O - N+1 queries, sync calls in async contexts
3. Serialization Overhead - Repeated parsing, unnecessary cloning
4. Scale-Breaking Assumptions - Code that won't survive increased load

When to Use

• "Why is this slow?"
• "Will this scale to 10K users?"
• "Review this code before production"
• "Find N+1 queries"

Common Patterns Detected

# N+1 Query
for item in items:
    fetch_related(item.id)  # Query per iteration

# Nested Loop O(n²)
for a in list_a:
    for b in list_b:
        compare(a, b)

# Blocking in Async
async def handler():
    requests.get(url)  # Blocks event loop

Output

### Critical Hotspots
| Location | Pattern | Complexity | Impact | Fix |

### Load-Sensitive Paths
| Path | Scaling Factor | Breaking Point | Symptoms |

### Scale-Breaking Assumptions
| Assumption | Current Safety | Failure Scenario | Risk |

Agent

Use subagent_type: performance-analyzer