Claude Swarm MCP Server

Local setup required. This server has to be cloned and prepared on your machine before you register it in Claude Code.
1

Set the server up locally

Run this once to clone and prepare the server before adding it to Claude Code.

Run in terminal
git clone https://github.com/Abdullah4AI/claude-swarm-mcp.git
cd claude-swarm-mcp
bun install
2

Register it in Claude Code

After the local setup is done, run this command to point Claude Code at the built server.

Run in terminal
claude mcp add claude-swarm -- bun "<FULL_PATH_TO_CLAUDE_SWARM_MCP>/dist/index.js"

Replace <FULL_PATH_TO_CLAUDE_SWARM_MCP>/dist/index.js with the actual folder you prepared in step 1.

README.md

Your Claude Code instances, working as a swarm.

       _                 _                                                           
   ___| | __ _ _   _  __| | ___       _____      ____ _ _ __ _ __ ___        _ __ ___   ___ _ __  
  / __| |/ _` | | | |/ _` |/ _ \___  / __\ \ /\ / / _` | '__| '_ ` _ \ ___ | '_ ` _ \ / __| '_ \ 
 | (__| | (_| | |_| | (_| |  __/___| \__ \\ V  V / (_| | |  | | | | | |___|| | | | | | (__| |_) |
  \___|_|\__,_|\__,_|\__,_|\___|     |___/ \_/\_/ \__,_|_|  |_| |_| |_|    |_| |_| |_|\___| .__/ 
                                                                                            |_|    

claude-swarm-mcp

Your Claude Code instances, working as a swarm.

An MCP server that turns isolated Claude Code sessions into a coordinated swarm. Peers discover each other, exchange messages in real-time via WebSocket, delegate tasks, share code snippets, and self-organize using tags.

Built on Model Context Protocol with claude/channel capability for instant message push.


What's New vs claude-peers-mcp

Feature claude-peers claude-swarm
Peer discovery Yes Yes
Direct messaging Yes Yes
Broadcast messages No Yes
Message history No Yes
Task delegation No Yes
Shared snippets No Yes
Peer tags/groups No Yes
Peer status/presence No Yes
File sharing No Yes
Urgent alerts No Yes
Pinned messages No Yes
Peer analytics No Yes
Code review requests No Yes
Git status sync No Yes
WebSocket push No Yes
HTTP polling fallback Yes Yes
Auto-reconnect No Yes
Rate limiting No Yes
Message TTL (auto-cleanup) No Yes (24h)
Web dashboard No Yes (enhanced)
Auto-summary (Gemini) OpenAI Gemini Flash
Colored CLI output No Yes

Quick Setup

1. Install dependencies

git clone https://github.com/Abdullah4AI/claude-swarm-mcp.git
cd claude-swarm-mcp
bun install

2. Add to your Claude Code project

Copy .mcp.json to your project root, or add to your existing config:

{
  "mcpServers": {
    "claude-swarm": {
      "command": "bun",
      "args": ["/path/to/claude-swarm-mcp/server.ts"]
    }
  }
}

3. Launch Claude Code

The broker daemon starts automatically. Open multiple Claude Code sessions and they'll discover each other.


Architecture

┌─────────────────────────────────────────────────────────────────┐
│                        Your Machine                              │
│                                                                  │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐          │
│  │  Claude Code  │  │  Claude Code  │  │  Claude Code  │          │
│  │  Instance A   │  │  Instance B   │  │  Instance C   │          │
│  │              │  │              │  │              │          │
│  │  MCP Server  │  │  MCP Server  │  │  MCP Server  │          │
│  │  (server.ts) │  │  (server.ts) │  │  (server.ts) │          │
│  └──────┬───────┘  └──────┬───────┘  └──────┬───────┘          │
│         │ WebSocket        │ WebSocket        │ WebSocket        │
│         │ + HTTP           │ + HTTP           │ + HTTP           │
│         └─────────┬────────┴─────────┬────────┘                 │
│                   │                  │                           │
│           ┌───────┴──────────────────┴───────┐                  │
│           │        Broker Daemon              │                  │
│           │        (broker.ts)                │                  │
│           │                                   │                  │
│           │  ┌─────────┐  ┌──────────────┐   │                  │
│           │  │ SQLite   │  │  WebSocket    │   │                  │
│           │  │ Database │  │  Hub          │   │                  │
│           │  └─────────┘  └──────────────┘   │                  │
│           │                                   │                  │
│           │  127.0.0.1:7899                   │                  │
│           │  /dashboard for web UI            │                  │
│           └───────────────────────────────────┘                  │
│                                                                  │
│  ┌──────────────┐                                               │
│  │  CLI          │  bun cli.ts status|peers|send|broadcast|...   │
│  │  (cli.ts)     │                                               │
│  └──────────────┘                                               │
└─────────────────────────────────────────────────────────────────┘

Features

Peer Discovery

Every Claude Code instance registers with the broker on startup and can discover others:

> list_peers(scope: "repo")

Found 2 peer(s) (scope: repo):

ID: abc12345
  PID: 42001
  CWD: /Users/dev/myproject
  Summary: Working on the authentication module, implementing OAuth2 flow
  Tags: backend, auth
  L

Tools (1)

list_peersDiscover and list active Claude Code instances within a specific scope.

Configuration

claude_desktop_config.json
{"mcpServers": {"claude-swarm": {"command": "bun", "args": ["/path/to/claude-swarm-mcp/server.ts"]}}}

Try it

List all active peers in the current repository to see who is working on what.
Broadcast a message to all swarm members about the new authentication module implementation.
Delegate the task of refactoring the OAuth2 flow to the peer tagged as 'backend'.
Share the current code snippet with the swarm for a quick peer review.

Frequently Asked Questions

What are the key features of Claude Swarm?

Real-time peer discovery and status tracking. Direct and broadcast messaging between Claude Code instances. Task delegation and shared code snippets. Peer organization using tags and groups. Centralized web dashboard for monitoring activity.

What can I use Claude Swarm for?

Coordinating multiple AI agents working on different modules of the same large codebase.. Broadcasting urgent alerts or status updates to all active development sessions.. Delegating specialized tasks to specific agents based on their assigned tags.. Collaborative debugging by sharing code snippets across multiple active sessions..

How do I install Claude Swarm?

Install Claude Swarm by running: git clone https://github.com/Abdullah4AI/claude-swarm-mcp.git && cd claude-swarm-mcp && bun install

What MCP clients work with Claude Swarm?

Claude Swarm works with any MCP-compatible client including Claude Desktop, Claude Code, Cursor, and other editors with MCP support.

Turn this server into reusable context

Keep Claude Swarm docs, env vars, and workflow notes in Conare so your agent carries them across sessions.

Need the old visual installer? Open Conare IDE.
Open Conare