moises-paschoalick/node-red-mcp-server

Developer tools

Эта страница переведена автоматически. Технические детали смотрите в оригинале. Открыть оригинал

11 stars 3 forks Качество 53 Тренд 8

Repository Install

Connects Node-RED to LLMs via the Model Context Protocol (MCP) for intelligent AI workflows.

Обзор

Connects Node-RED to LLMs via the Model Context Protocol (MCP) for intelligent AI workflows. - Node-RED configured with MCP (Model Context Protocol) for LLM integration This project was originally designed with a Docker-based setup using two containers: Node-RED and an mcp-host responsible for running the MCP (Model Context Protocol) server. This architecture enabled Node-RED flows to interact with LLMs through a dedicated MCP layer. With the release of node-red-contrib-mcp-tools version , the architecture has been significantly simplified. To enable Streamable HTTP transport, just install the node-red-contrib-mcp-tools component — no extra containers or services needed. You can run both local and remote MCP servers. Bearer Token: GO https://dashboard.exa.ai/api-keys and generate one API Keys The repo contains a ready-to-use Railway configuration:

README

Показан фрагмент оригинального README. Открыть оригинал.

Node-RED MCP Server

Connects Node-RED to LLMs via the Model Context Protocol (MCP) for intelligent AI workflows.

🚀 Ready-to-deploy Railway template - Node-RED configured with MCP (Model Context Protocol) for LLM integration

📋 Project Description

This project was originally designed with a Docker-based setup using two containers: Node-RED and an mcp-host responsible for running the MCP (Model Context Protocol) server. This architecture enabled Node-RED flows to interact with LLMs through a dedicated MCP layer.

With the release of node-red-contrib-mcp-tools version 2.2.0, the architecture has been significantly simplified.

To enable Streamable HTTP transport, just install the node-red-contrib-mcp-tools component — no extra containers or services needed.

You can run both local and remote MCP servers.

Example of a remote MCP:

MCP Server URL:

https://mcp.exa.ai/mcp

Bearer Token: GO https://dashboard.exa.ai/api-keys and generate one API Keys

a98e0***************************...

🔗 Find other MCPs on smithery.ai

🚀 Deploy on Railway

How it works

The repo contains a ready-to-use Railway configuration:

File	Purpose
`Dockerfile.railway`	Build image from repo root — copies flows, MCP servers, and provisioning script
`railway.toml`	Tells Railway which Dockerfile to use and sets the start command
`docker-init/user-provision.js`	Runs on every container start — hashes the admin password and writes it to `settings.js`

Build flow:

repo root (build context)
  ├── node-red-docker/node_red_data  →  /data          (flows, settings, nodes)
  ├── mcp-server/v1                  →  /data/mcp-server/v1  (MCP server + gdrive)
  └── docker-init/user-provision.js  →  /data/user-provision.js

Start command (from railway.toml):

cd /data && node user-provision.js && node-red --userDir /data

user-provision.js reads ADMIN_PASSWORD env var (or generates a random one), bcrypt-hashes it, and writes it into settings.js before Node-RED starts.