ContextQMD
Back to Ruflo

Claude Flow Benchmark Examples

v2/benchmark/examples/README.md

3.6.306.0 KB
Original Source

Claude Flow Benchmark Examples

This directory contains organized examples for using the Claude Flow benchmark suite with different strategies, coordination modes, and real-world scenarios.

Directory Structure

examples/
├── basic/                  # Simple examples for getting started
├── advanced/               # Complex examples with advanced features
├── real/                   # Real claude-flow execution examples
├── cli/                    # Command-line interface examples
└── output/                 # Generated results and metrics

Basic Examples (basic/)

Getting started with simple benchmarks:

  • simple_swarm.py - Basic swarm coordination benchmark
  • simple_hive_mind.py - Basic hive-mind collective intelligence
  • simple_sparc.py - Basic SPARC methodology (TDD approach)
  • claude_optimizer_example.py - Claude optimizer usage
  • example_usage.py - General usage patterns

Run a basic example:

bash
cd basic/
python3 simple_swarm.py

Advanced Examples (advanced/)

Complex benchmarks with advanced features:

  • parallel_benchmarks.py - Concurrent execution strategies
  • optimization_suite.py - Performance tuning and efficiency analysis
  • comparative_analysis.py - Multi-strategy comparison
  • demo_comprehensive.py - Comprehensive feature demonstration
  • parallel_benchmark_demo.py - Parallel execution patterns

Run an advanced example:

bash
cd advanced/
python3 parallel_benchmarks.py

Real Examples (real/)

Production-ready benchmarks with actual claude-flow execution:

  • real_swarm_benchmark.py - Real swarm execution with comprehensive metrics
  • real_token_tracking.py - Token consumption analysis and cost optimization
  • real_performance.py - System performance monitoring and analysis
  • real_hive_mind_benchmark.py - Real hive-mind collective intelligence
  • real_sparc_benchmark.py - Real SPARC methodology execution
  • real_benchmark_examples.py - Various real benchmark scenarios

Run a real example:

bash
cd real/
python3 real_swarm_benchmark.py

CLI Examples (cli/)

Command-line interface demonstrations:

  • cli_examples.sh - Comprehensive CLI usage examples
  • batch_benchmarks.sh - Batch execution scripts

Run CLI examples:

bash
cd cli/
./cli_examples.sh

Run batch benchmarks:

bash
cd cli/
./batch_benchmarks.sh

Quick Start

1. Simple Swarm Benchmark

bash
python3 basic/simple_swarm.py

2. Real Performance Analysis

bash
python3 real/real_performance.py

3. Token Tracking and Cost Analysis

bash
python3 real/real_token_tracking.py

4. Comprehensive CLI Demo

bash
./cli/cli_examples.sh

Example Types by Use Case

Learning and Testing

  • basic/simple_*.py - Start here for learning
  • cli/cli_examples.sh - Command-line reference

Development and Optimization

  • advanced/optimization_suite.py - Performance tuning
  • real/real_performance.py - System monitoring
  • real/real_token_tracking.py - Cost optimization

Production Assessment

  • real/real_swarm_benchmark.py - Production readiness
  • advanced/comparative_analysis.py - Strategy comparison
  • cli/batch_benchmarks.sh - Automated testing

Research and Analysis

  • advanced/parallel_benchmarks.py - Concurrent execution
  • real/real_hive_mind_benchmark.py - Collective intelligence
  • advanced/comparative_analysis.py - Multi-methodology comparison

Output and Results

All examples save results to the output/ directory with timestamps:

output/
├── simple_swarm_metrics.json
├── parallel_benchmark_results.json
├── token_tracking_metrics_*.json
├── performance_analysis_*.json
└── batch_results_*/

Requirements

Python Dependencies:

bash
pip install psutil  # For system monitoring

Claude Flow:

bash
npm install -g claude-flow@alpha

Benchmark Suite:

bash
pip install -e .  # From benchmark root directory

Configuration

Most examples can be configured by modifying parameters at the top of each script:

python
# Example configuration
config = {
    "agents": 5,
    "coordination": "hierarchical",
    "strategy": "development",
    "timeout": 180
}

Best Practices

  1. Start Simple: Begin with basic/ examples
  2. Monitor Resources: Use real/real_performance.py for system analysis
  3. Track Costs: Use real/real_token_tracking.py for cost optimization
  4. Compare Strategies: Use advanced/comparative_analysis.py
  5. Automate Testing: Use cli/batch_benchmarks.sh

Integration with CI/CD

Use batch scripts for automated testing:

yaml
# GitHub Actions example
- name: Run Benchmark Suite
  run: |
    cd benchmark/examples/cli/
    ./batch_benchmarks.sh
    
- name: Upload Results
  uses: actions/upload-artifact@v3
  with:
    name: benchmark-results
    path: benchmark/examples/output/

Troubleshooting

Common Issues:

  1. Command not found: Ensure claude-flow@alpha is installed globally
  2. Permission denied: Run chmod +x cli/*.sh
  3. Import errors: Install with pip install -e . from benchmark root
  4. Timeout errors: Increase timeout values in configurations

Debug Mode:

Add --debug flag to commands for verbose output:

bash
python3 real/real_swarm_benchmark.py --debug

Contributing

To add new examples:

  1. Choose appropriate directory (basic/, advanced/, real/, cli/)
  2. Follow naming convention: {purpose}_{type}_{description}.py
  3. Include comprehensive docstrings and comments
  4. Save outputs to output/ directory with timestamps
  5. Update this README with usage instructions

Support

  • Documentation: /workspaces/claude-code-flow/benchmark/docs/
  • Issues: Create GitHub issues with example logs
  • Community: Join discussions in GitHub Discussions

Next Steps:

  1. Try basic examples to understand concepts
  2. Run real examples for production insights
  3. Use CLI examples for automation
  4. Customize advanced examples for specific needs