Claude-Force Documentation

Production-Ready Multi-Agent Orchestration System for Claude

Welcome to the comprehensive documentation for claude-force, a powerful framework for orchestrating multiple Claude AI agents to solve complex software engineering tasks.

Quick Links

Installation Guide - Get started in 30 seconds
Quick Start - Build your first multi-agent workflow
API Reference - Complete API documentation
Guides - In-depth tutorials and best practices
Examples - Real-world usage examples

What is Claude-Force?

Claude-Force is a Python framework that enables you to:

Orchestrate Multiple AI Agents: Coordinate specialized Claude agents for different tasks
Semantic Agent Selection: Automatically recommend the best agent for your task using embeddings
Hybrid Model Orchestration: Optimize costs by automatically selecting Haiku/Sonnet/Opus based on task complexity
Performance Tracking: Monitor execution time, token usage, and costs
Workflow Composition: Create custom multi-agent workflows
Marketplace Integration: Install and share community agents

Key Features

🧠 Semantic Agent Recommendation

claude-force recommend --task "Review authentication code for security"
# → Recommends: security-specialist (confidence: 0.92)

⚡ Hybrid Model Orchestration

Automatically selects the optimal Claude model based on task complexity:

Haiku for simple tasks (typo fixes, formatting)
Sonnet for moderate tasks (code reviews, refactoring)
Opus for complex tasks (architecture design, security audits)

📊 Performance Analytics

claude-force analyze performance
# Shows: execution times, token usage, costs, success rates

🎨 Project Initialization

claude-force init ./my-project \
  --description "Build a chat application" \
  --name "chat-app"
# Creates optimized .claude/ configuration

Installation

From PyPI (Recommended)

pip install claude-force
claude-force --help

From Source

git clone https://github.com/khanh-vu/claude-force.git
cd claude-force
pip install -e .

Quick Example

from claude_force.orchestrator import AgentOrchestrator

# Initialize orchestrator
orchestrator = AgentOrchestrator(
    config_path=".claude/claude.json",
    enable_tracking=True
)

# Run single agent
result = orchestrator.run_agent(
    agent_name="code-reviewer",
    task="Review this authentication function for security issues"
)

print(result.output)

# Run multi-agent workflow
results = orchestrator.run_workflow(
    workflow_name="full-stack-feature",
    task="Implement user authentication with JWT"
)

for result in results:
    print(f"{result.agent_name}: {result.success}")

Architecture

Claude-Force consists of several key components:

AgentOrchestrator: Core execution engine
SemanticAgentSelector: AI-powered agent recommendation
HybridOrchestrator: Model selection and cost optimization
PerformanceTracker: Metrics collection and analytics
WorkflowComposer: Custom workflow creation
AgentMarketplace: Community agent sharing

Documentation Sections

For Users

Installation: System requirements and installation methods
Quick Start: Get started in minutes
CLI Reference: All command-line commands
Guides: Tutorials and best practices

For Developers

API Reference: Complete Python API documentation
Architecture: System design and internals
Contributing: How to contribute to claude-force
Testing: Running and writing tests

Advanced Topics

Semantic Selection: How agent recommendation works
Performance Optimization: Cost and latency optimization
Custom Workflows: Building multi-agent workflows
Marketplace: Publishing and installing agents

System Requirements

Python 3.8 or higher (3.10+ recommended)
Anthropic API key (Get one here)
100MB disk space

Optional Dependencies

sentence-transformers: For semantic agent selection (recommended)
numpy: For vector operations (required if using semantic selection)

Support

GitHub Issues: https://github.com/khanh-vu/claude-force/issues
Documentation: https://claude-force.readthedocs.io
Source Code: https://github.com/khanh-vu/claude-force

License

MIT License - see LICENSE file for details.

Next Steps

New Users: Start with the Quick Start Guide
Developers: Check out the API Reference
Advanced Users: Explore Guides for in-depth topics

Version: 2.1.0 Last Updated: 2025-11-14 Status: Production Ready