Claude Swarm

Provides a local MCP server that enables persistent, coordinated Claude AI agents for multi-agent workflows.

python

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "mayank1805-claude_swarm_mcp_agent": {
      "command": "python3",
      "args": [
        "/path/to/claude-swarm-mcp/claude_swarm_mcp_server.py"
      ],
      "env": {
        "ANTHROPIC_API_KEY": "YOUR_API_KEY_PLACEHOLDER",
        "CLAUDE_SWARM_DEBUG": "true",
        "CLAUDE_SWARM_STORAGE_DIR": "/Users/you/claude_agents"
      }
    }
  }
}

You can run and manage a local MCP server that coordinates multiple Claude AI agents. This server uses the Model Context Protocol to enable persistent, interconnected agents that operate within Claude Desktop, saving conversations locally and enabling intelligent handoffs between specialized roles for complex tasks like financial analysis, customer service, and research.

How to use

Install and start the MCP server locally, then configure your Claude Desktop client to connect to it. You will create, manage, and chat with specialized agents, and you can trigger multi-agent workflows such as a complete financial analysis or a customer support team coordination. The server persists agents and conversations on your machine and coordinates handoffs between agents based on expertise.

How to install

Prerequisites you need before installing the MCP server are Python 3.10 or newer, Claude Desktop installed, and an Anthropic API key with billing enabled.

git clone https://github.com/yourusername/claude-swarm-mcp.git
cd claude-swarm-mcp

pip install -r requirements.txt

Configure Claude Desktop to point at the local MCP server. Use the following configuration, which runs the MCP server locally via Python and passes your Anthropic API key through the environment.

{
  "mcpServers": {
    "claude_swarm": {
      "command": "python3",
      "args": ["/path/to/claude-swarm-mcp/claude_swarm_mcp_server.py"],
      "env": {
        "ANTHROPIC_API_KEY": "your-api-key-here"
      }
    }
  }
}

Start the MCP server from the command line. This runs the server locally so Claude Desktop can connect and coordinate agents.

python3 claude_swarm_mcp_server.py

Additional content and notes

After starting, restart Claude Desktop if needed and test a workflow, for example creating a finance team with a company name. The server will persist agents and conversations locally, and agents will coordinate to provide comprehensive analyses.

Available tools

create_agent

Create a new specialized agent with given name and instructions to perform a focused task.

list_agents

List all saved agents with their statuses and summaries.

chat_with_agent

Interact with a specific agent or set of agents to perform tasks and receive responses.

delete_agent

Remove an agent permanently from storage along with its conversation history.

create_finance_team

Generate a complete financial analysis team with predefined specialized roles.

get_conversation_history

Retrieve chat history and agent handoffs for auditing or review.

reset_conversation

Clear the current conversation history for a fresh start.

Built by

VeilStrat

AI signals for GTM teams