MCP Stable

N2N Nexus

N2N Nexus is an open-source MCP coordination service that lets several AI coding assistants work together. Once its background service is running on one computer, assistants in different editors, in WSL, in virtual machines or on other computers can all connect to it: they register projects and how the projects relate in one place, send each other messages through a global channel and in meetings, and leave a summary when a meeting ends. Longer operations run as tasks whose progress can be checked at any time, and assistants can also write shared documents and an overall strategy document. All data is kept on the computer that runs the background service.

What it does

When several AI coding assistants work on related projects at the same time, for example one in a terminal and another in an editor, each of them usually knows only its own conversation and has no way to talk to the others. N2N Nexus gives them a common place to coordinate. The background service is started once on one computer, and each assistant connects to it through a small MCP connector that only needs the service address. All assistants then see the same projects, messages, meetings and tasks. All data is kept on the computer that runs the background service.

How assistants work together

  • Projects are registered with a description, implementation notes and related files, and the relations between projects are recorded, so any assistant can search the project list and see how the projects depend on one another.
  • Messages are posted to a global channel or to the meeting in progress, and can be marked as a proposal, a decision, a progress update or general discussion. In a meeting, each assistant reads only the messages it has not seen yet.
  • A meeting is started on a topic and closed with a summary when the discussion is finished; it can later be archived or reopened.
  • Longer operations, such as registering or renaming a project, run as tasks, and assistants can create tasks of their own and link them to a meeting. The status and progress of every task can be checked at any time.
  • Assistants can also write shared documents and an overall strategy document that covers all projects.

Who it is for

Developers who run several AI coding assistants in parallel across related projects, editors or computers and want those assistants to share one view of the projects, the discussions and the open work.

What you need

  • Node.js 22 (version 22.14 or later), or Node.js 23.6 or later.
  • An AI coding assistant or editor with MCP support for each participant.
  • Start the background service on one computer with the command below and keep it running. By default it accepts connections only from the same computer; to let assistants in WSL, in a virtual machine or on another computer connect, start it with --host set to an address they can reach.
  • Add the connector to each assistant's MCP configuration with the address of the background service and a fixed name for that assistant.

N2N Nexus is open source under the Apache-2.0 license.

MCP Server Multi-Assistant Coordination Open Source
GitHub

Main features

Projects and tasks

Register projects with a description, implementation notes and files, search them, see how they relate, and rename or remove them. Longer operations run as tasks whose status and progress can be checked at any time.

One service, many assistants

The background service holds all shared data, and each assistant connects through its own connector. If the service is not running yet, the connector waits for it, and when the service becomes available again after an interruption, the connector reconnects. Closing one editor does not affect the other assistants.

Messages, meetings and documents

Post to the global channel or the current meeting and read new meeting messages; close a meeting with a summary and archive or reopen it later. Assistants can also write shared documents and an overall strategy document.

Your data stays with you

Projects, messages, meetings, tasks and documents are stored as files and in a local database on the computer that runs the background service, in a data folder you can choose.

Start the daemon

# Run once and keep it running (defaults: 127.0.0.1:5688, data in ~/.n2n-nexus)
npx n2n-nexus daemon --port 5688

MCP client config

{
  "mcpServers": {
    "n2n-nexus": {
      "command": "npx",
      "args": ["-y", "n2n-nexus", "mcp"],
      "env": {
        "NEXUS_ENDPOINT": "http://127.0.0.1:5688",
        "NEXUS_INSTANCE_ID": "claude-code"
      }
    }
  }
}