AgentSkillsCN

Claude Flow Performance

性能剖析、基准测试、优化建议、瓶颈检测，以及指标追踪。适用于进行性能基准测试、剖析应用、识别瓶颈，或优化代理工作流时使用。

SKILL.md
--- frontmatter
name: "Claude Flow Performance"
description: "Performance profiling, benchmarking, optimization recommendations, bottleneck detection, and metrics tracking. Use when benchmarking performance, profiling applications, identifying bottlenecks, or optimizing agent workflows."

Claude Flow Performance

Performance module providing benchmarking, Flash Attention validation, profiling, bottleneck detection, optimization recommendations, and metrics tracking.

Quick Command Reference

TaskCommand
Run benchmarknpx @claude-flow/cli@latest performance benchmark
Profile appnpx @claude-flow/cli@latest performance profile
View metricsnpx @claude-flow/cli@latest performance metrics
Optimizenpx @claude-flow/cli@latest performance optimize
Find bottlenecksnpx @claude-flow/cli@latest performance bottleneck

Core Commands

performance benchmark

Run performance benchmarks.

bash
npx @claude-flow/cli@latest performance benchmark

performance profile

Profile application performance.

bash
npx @claude-flow/cli@latest performance profile

performance metrics

View and export performance metrics.

bash
npx @claude-flow/cli@latest performance metrics

performance optimize

Run performance optimization recommendations.

bash
npx @claude-flow/cli@latest performance optimize

performance bottleneck

Identify performance bottlenecks.

bash
npx @claude-flow/cli@latest performance bottleneck

Common Patterns

Full Performance Analysis

bash
# Run benchmarks
npx @claude-flow/cli@latest performance benchmark

# Profile
npx @claude-flow/cli@latest performance profile

# Find bottlenecks
npx @claude-flow/cli@latest performance bottleneck

# Get optimization suggestions
npx @claude-flow/cli@latest performance optimize

Monitor Performance Over Time

bash
# Collect metrics
npx @claude-flow/cli@latest performance metrics

# Benchmark after changes
npx @claude-flow/cli@latest performance benchmark

Key Options

  • •--verbose: Enable verbose output
  • •--format: Output format (text, json, table)

Programmatic API

typescript
import { Benchmarker, Profiler, MetricsCollector } from '@claude-flow/performance';

// Benchmark
const bench = new Benchmarker();
const results = await bench.run();

// Profile
const profiler = new Profiler();
const profile = await profiler.profile(targetFn);

// Collect metrics
const metrics = new MetricsCollector();
const data = await metrics.collect();

RAN DDD Context

Bounded Context: Cross-Cutting Related Skills: claude-flow, claude-flow-neural

References