MCP-Swarm

MCP Swarm is a Node.js-based platform that coordinates multiple AI agents (such as Claude, Cursor, Windsurf, and OpenCode) into a single collaborative workflow. It provides a centralized orchestration layer, real-time collaboration features, and a suite of 35 smart tools organized into categories like core system, task management, file and git collaboration, chat and governance, analytics, intelligence, and infra. The system handles task assignment, locking to prevent conflicts, and persistent swarm history in the swarm/ directory so agents can resume work seamlessly. To use it, install the MCP Swarm package, run the server, and interact with the provided tools through its command surface and API to create tasks, manage files, broadcast chats, perform analyses, and index knowledge.

Prerequisites:

• Node.js 18+ and npm installed
• Access to a terminal/command prompt

Step 1: Install the MCP Swarm globally (recommended for quick usage)

• npm install -g mcp-swarm@latest

Step 2: Run the MCP Swarm server

• mcp-swarm

Step 3: Verify the server is running

• Open http://localhost:37373 (or the URL printed in the console) to view the live dashboard

Step 4: Optional configuration

• If you need to customize environment variables or ports, create a config file or set environment variables as needed before starting the server. See the documentation for advanced setup and plugin loading options.

Tips and common issues:

• Ensure Node.js 18+ is installed to avoid compatibility issues with the latest MCP Swarm features.
• If the server fails to start due to port conflicts, override the port with an environment variable or CLI flag as described in the docs.
• The swarm/ directory stores persistent state; ensure it is writable by the running process.
• When using the 35 tools, familiarize yourself with the key actions for each tool (e.g., swarm_task: create/list/update, swarm_chat: broadcast/dashboard/thought, swarm_vector: init/add/search).
• For upgrading, use npm install -g mcp-swarm@latest to get the latest features and fixes.
• If you plan to run in a production environment, consider enabling persistence, backups, and access controls for the dashboard and API endpoints.