N2N Nexus — Local-First MCP Coordination Hub

N2N Nexus

A local MCP coordination hub for multiple AI coding assistants working around the same project state.

n2n-nexus is for teams that no longer have just one AI assistant in one editor. It runs a local daemon as the shared coordination layer, then lets lightweight MCP adapters from different IDEs, shells, or machines connect to the same state.

It works as a coordination hub, not a memory replacement. It owns project registry data, meetings, messages, tasks, shared docs, and assets. The value is that assistants can register the active project, exchange structured updates, keep meeting history, and coordinate async work without every client needing broad filesystem or backend access. Network exposure and stored project materials remain operational decisions for the user or team.

MCP Server Coordination Daemon Multi-Agent

Coordination Layer

Task Coordination

Create, poll, and coordinate async work across multiple connected assistants.

Shared Local Daemon

Run one always-on daemon as the coordination source of truth.

Meetings and Messages

Keep structured assistant discussion, decisions, proposals, and updates in one local workspace.

Adapter Boundary

Use stateless MCP adapters so IDE clients connect without each owning the data store.

Daemon + MCP setup

# Start the daemon
npx n2n-nexus daemon --port 5688

# Point MCP adapters at it
NEXUS_ENDPOINT=http://127.0.0.1:5688 npx -y n2n-nexus mcp

MCP client config

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