🧠 Claude-Flow: Advanced AI Agent Orchestration System for Claude Code

A sophisticated multi-agent orchestration system built with Deno and TypeScript, featuring advanced coordination, memory management, and terminal integration with 100% test coverage.

Features

• Multi-Agent Orchestration: Spawn and manage multiple Claude agents with different roles and capabilities
• Task Scheduling: Advanced task queue with priority-based scheduling and dependency management
• Memory Management: Persistent memory storage with hybrid SQLite/Markdown backends
• Terminal Management: Efficient terminal pooling and session management
• Resource Coordination: Deadlock detection and prevention for shared resources
• MCP Support: Model Context Protocol server for tool integration
• Interactive CLI: Rich command-line interface with REPL mode

Installation

Using NPM (recommended)

npx claude-flow

Or install globally:

npm install -g claude-flow

Using Deno

deno install --allow-all --name claude-flow https://raw.githubusercontent.com/ruvnet/claude-code-flow/main/src/cli/index.ts

From Source

git clone https://github.com/ruvnet/claude-code-flow.git
cd claude-flow
deno task install

Quick Start

1. Initialize configuration:

   claude-flow config init

2. Start the orchestrator:

   claude-flow start

3. Spawn an agent:

   claude-flow agent spawn researcher --name "Research Assistant"

4. Create a task:

   claude-flow task create research "Find information about quantum computing"

Architecture

Claude-Flow uses a modular architecture with the following components:

• Orchestrator: Central coordinator managing all system components
• Terminal Manager: Handles terminal sessions with pooling and recycling
• Memory Manager: Persistent storage with caching and indexing
• Coordination Manager: Task scheduling and resource management
• MCP Server: Tool integration via Model Context Protocol

Configuration

Default configuration file (claude-flow.config.json):

{
  "orchestrator": {
    "maxConcurrentAgents": 10,
    "taskQueueSize": 100,
    "healthCheckInterval": 30000,
    "shutdownTimeout": 30000
  },
  "terminal": {
    "type": "auto",
    "poolSize": 5,
    "recycleAfter": 10,
    "healthCheckInterval": 60000,
    "commandTimeout": 300000
  },
  "memory": {
    "backend": "hybrid",
    "cacheSizeMB": 100,
    "syncInterval": 5000,
    "conflictResolution": "crdt",
    "retentionDays": 30
  },
  "coordination": {
    "maxRetries": 3,
    "retryDelay": 1000,
    "deadlockDetection": true,
    "resourceTimeout": 60000,
    "messageTimeout": 30000
  },
  "mcp": {
    "transport": "stdio",
    "port": 3000,
    "tlsEnabled": false
  },
  "logging": {
    "level": "info",
    "format": "json",
    "destination": "console"
  }
}

Agent Types

Claude-Flow supports multiple agent types:

• Coordinator: Plans and delegates tasks to other agents
• Researcher: Gathers and analyzes information
• Implementer: Writes code and creates solutions
• Analyst: Identifies patterns and generates insights
• Custom: User-defined agent types

CLI Commands

Global Options

• -c, --config <path>: Path to configuration file
• -v, --verbose: Enable verbose logging
• --log-level <level>: Set log level (debug, info, warn, error)

Commands

start

Start the Claude-Flow orchestration system

claude-flow start [options]
  -d, --daemon              Run as daemon in background
  -p, --port <port>         MCP server port
  --mcp-transport <type>    MCP transport type (stdio, http)

agent

Manage agents

claude-flow agent <subcommand>
  spawn <type>              Spawn a new agent
  list                      List all agents
  terminate <agent-id>      Terminate an agent
  info <agent-id>          Get agent information

task

Manage tasks

claude-flow task <subcommand>
  create <type> <desc>      Create a new task
  list                      List all tasks
  status <task-id>          Get task status
  cancel <task-id>          Cancel a task
  workflow <file>           Execute workflow from file

memory

Manage agent memory

claude-flow memory <subcommand>
  query                     Query memory entries
  export <file>             Export memory to file
  import <file>             Import memory from file
  stats                     Show memory statistics
  cleanup                   Clean up old entries

config

Manage configuration

claude-flow config <subcommand>
  show                      Show current configuration
  get <path>                Get specific config value
  set <path> <value>        Set config value
  init [file]               Initialize config file
  validate <file>           Validate config file

repl

Start interactive REPL mode

claude-flow repl

Workflow Example

Create a workflow file (example-workflow.json):

{
  "name": "Research and Analysis Workflow",
  "tasks": [
    {
      "id": "research-1",
      "type": "research",
      "description": "Research quantum computing basics",
      "assignTo": "researcher"
    },
    {
      "id": "analyze-1",
      "type": "analysis",
      "description": "Analyze research findings",
      "dependencies": ["research-1"],
      "assignTo": "analyst"
    },
    {
      "id": "report-1",
      "type": "report",
      "description": "Generate summary report",
      "dependencies": ["analyze-1"],
      "assignTo": "coordinator"
    }
  ]
}

Execute the workflow:

claude-flow task workflow example-workflow.json

Development

Prerequisites

• Deno 1.38+ or Node.js 16+
• Git

Setup

git clone https://github.com/ruvnet/claude-code-flow.git
cd claude-flow
deno task dev

Testing

deno task test

Building

deno task build

API Usage

Claude-Flow can also be used programmatically:

import { Orchestrator } from 'claude-flow';

const orchestrator = new Orchestrator(config);
await orchestrator.initialize();

// Spawn an agent
const sessionId = await orchestrator.spawnAgent({
  id: 'agent-1',
  name: 'Research Agent',
  type: 'researcher',
  // ... other properties
});

// Create and assign a task
await orchestrator.assignTask({
  id: 'task-1',
  type: 'research',
  description: 'Research AI trends',
  // ... other properties
});

Contributing

Contributions are welcome! Please read our Contributing Guide for details.

License

MIT License - see LICENSE file for details.

Support

• Documentation: https://claude-flow.dev/docs
• Issues: GitHub Issues
• Discussions: GitHub Discussions

Roadmap

• Web UI for visual orchestration
• Plugin system for custom agent types
• Distributed orchestration support
• Enhanced monitoring and analytics
• Integration with more AI models
• Workflow templates library

---

Built with ❤️ by rUv for the Claude community