Claude Code Integration

Use Codeflash as a Claude Code subagent for seamless AI-powered code optimization

Claude Code Integration

Codeflash integrates natively with Claude Code through specialized subagents - dedicated AI assistants that provide comprehensive Python performance optimization capabilities directly within your development workflow.

🚀 Quick Setup

1. Install Codeflash

# Install codeflash (includes Claude Code integration)
pip install codeflash

2. Integrate with Claude Code

# Automatically set up native subagents
codeflash integrate claude

3. Restart Claude Code

After setup, restart Claude Code to load the new MCP server.

✨ Three Specialized Subagents

Codeflash provides three dedicated subagents, each with specialized expertise:

🚀 @codeflash-optimizer - AI-Powered Optimization

The primary optimization specialist that implements performance improvements:

"@codeflash-optimizer optimize the bubble_sort function"
"@codeflash-optimizer trace my script and optimize bottlenecks"
"@codeflash-optimizer set up GitHub Actions for continuous optimization"

Capabilities:

• Function and file-level optimization (2-55x speedups)
• End-to-end script tracing and optimization
• Algorithm improvements and efficiency gains
• Project setup and CI/CD integration

🔍 @codeflash-profiler - Performance Analysis

Expert in identifying bottlenecks and analyzing performance patterns:

"@codeflash-profiler analyze why my script is slow"
"@codeflash-profiler profile this function and find bottlenecks"
"@codeflash-profiler benchmark my code and show performance metrics"

Capabilities:

• Execution tracing and hotspot identification
• Performance benchmarking and statistical analysis
• Memory usage analysis and optimization recommendations
• Scalability assessment and bottleneck prioritization

👁️ @codeflash-reviewer - Performance Code Reviews

Specialized in reviewing code for performance issues during development:

"@codeflash-reviewer review this code for performance issues"
"@codeflash-reviewer check my pull request for optimization opportunities"
"@codeflash-reviewer analyze the algorithmic complexity of this function"

Capabilities:

• Algorithmic efficiency analysis (time/space complexity)
• Performance anti-pattern detection
• Library usage optimization recommendations
• Code review integration with performance focus

🔧 Available MCP Tools

The integration provides these tools that Claude Code can use:

Tool	Description
optimize_function	Optimize a specific function by name
optimize_file	Optimize all functions in a file
trace_and_optimize	Trace script execution and optimize
optimize_from_replay_tests	Use test files to guide optimization
optimize_all_functions	Optimize entire project (use with caution)
initialize_project	Set up codeflash in a project
setup_github_actions	Configure CI/CD optimization
verify_installation	Test that codeflash works correctly
run_benchmarks	Execute performance benchmarks
get_codeflash_status	Check current project setup
get_optimization_help	Get usage help and best practices

📋 Usage Examples

Basic Optimization

Human: I have a slow sorting function in my project. Can you help optimize it?

Claude: I'll help you optimize your sorting function using codeflash. Let me first check what functions are available in your project and then optimize the sorting function.

[Claude uses codeflash tools to find and optimize the function]

End-to-End Script Optimization

Human: My data processing script takes too long. Can you trace it and find optimizations?

Claude: I'll use codeflash to trace your script execution and identify optimization opportunities based on real usage patterns.

[Claude uses trace_and_optimize tool]

Project Setup

Human: I want to set up codeflash in my new Python project

Claude: I'll help you initialize codeflash in your project and set up the optimal configuration.

[Claude uses initialize_project and setup tools]

⚙️ Configuration

Manual Setup

If automatic setup doesn’t work, you can manually add codeflash to your Claude Code configuration:

1. Find your Claude Code config file:

   • macOS: ~/.claude/config.json or ~/Library/Application Support/Claude Code/config.json
   • Linux: ~/.config/claude/config.json
   • Windows: %APPDATA%/Claude Code/config.json

2. Add the MCP server configuration:

{
  "mcpServers": {
    "codeflash": {
      "command": "codeflash-mcp",
      "args": [],
      "env": {},
      "disabled": false
    }
  }
}

Verification

Check if the integration is working:

# Check integration status
codeflash setup status

# Test in Claude Code
# Ask: "Show me codeflash help" or "What codeflash tools are available?"

🔍 Troubleshooting

Common Issues

”codeflash-mcp command not found"

# Reinstall with MCP dependencies
pip uninstall codeflash
pip install codeflash[mcp]

"MCP server failed to start"

# Check if dependencies are installed
python -c "import mcp, fastmcp; print('MCP dependencies OK')"

# Test the MCP server directly
codeflash-mcp

"Claude Code doesn’t recognize codeflash”

1. Ensure Claude Code is restarted after setup
2. Check config file exists and is valid JSON
3. Verify MCP server path is correct

Debug Commands

# Show detailed integration status
codeflash setup status

# Remove and reconfigure
codeflash setup claude-code --remove
codeflash setup claude-code --force

# Test MCP server manually
python -m codeflash.mcp.server

🎯 Best Practices

Optimization Workflow

1. Start Small: Begin with single function optimization
2. Use Tracing: For end-to-end scripts, use trace-based optimization
3. Review Changes: Always review optimizations before merging
4. Benchmark: Use benchmarking to measure improvements
5. Iterative: Optimize iteratively rather than all at once

Claude Code Tips

• Be specific about which functions/files to optimize
• Ask for explanations of optimizations made
• Request benchmarking to validate improvements
• Use natural language - Claude Code understands context

Performance Considerations

• Large project optimization can take significant time
• Use --no-pr flag for local-only testing
• Set up GitHub Actions for continuous optimization

🚀 Advanced Usage

Custom Workflows

"Create a codeflash optimization workflow for my data pipeline project that:
1. Traces the main processing script
2. Optimizes the top 5 slowest functions  
3. Runs benchmarks to verify improvements
4. Creates a PR with the optimizations"

Integration with CI/CD

"Help me set up automated codeflash optimization in my GitHub Actions workflow"

Performance Analysis

"Analyze my Python project's performance bottlenecks and suggest which functions to optimize first"

📚 Learn More

• Codeflash Documentation
• Model Context Protocol
• Claude Code Documentation

---

Codeflash Claude Code integration brings AI-powered code optimization directly into your development workflow, making performance improvements as easy as asking for them.