Overview
What is MCP Server + n8n?
This application combines the n8n workflow automation platform with an MCP Server using Docker to create an environment for experimentation. It allows users to test MCP Server features through the visual interface of n8n. The project is intended for educational and experimental purposes.
How to use MCP Server + n8n?
Install Docker and Docker Compose, then clone the repository and run docker-compose up -d. Access n8n at http://localhost:5678, create an account, and build a workflow. Add an AI Agent node, configure a Chat Model (e.g., OpenAI or Anthropic), set up a Simple Memory, and attach an MCP Client Tool with the SSE endpoint http://mcp_server:8000/sse. Finally, test the tools by opening the chat and sending commands like ping or acho eu.
Key features of MCP Server + n8n
- Docker-based setup for quick installation and configuration.
- Visual integration of n8n workflows with MCP Server tools.
- Persistent data storage in
n8n_dataandmcp_serverdirectories. - Pre‑defined MCP tools:
pingandecho. - Support for multiple LLM providers (OpenAI, Anthropic).
- Ability to modify the MCP server code in
server.py.
Use cases of MCP Server + n8n
- Experimenting with the MCP protocol in a visual workflow environment.
- Learning how to connect AI agents to external tools via MCP.
- Prototyping and testing custom MCP tools.
- Automating tasks that call MCP tools through n8n workflows.
- Demonstrating MCP capabilities in educational settings.
FAQ from MCP Server + n8n
What are the dependencies for MCP Server + n8n?
Docker and Docker Compose are required. On Windows, Docker Desktop and Git are needed; on Linux, docker.io, docker-compose, and docker-buildx.
Where does MCP Server + n8n store data?
Data is kept in persistent directories: n8n_data holds workflows, credentials, and configuration; mcp_server contains the MCP server Python code and its virtual environment.
What transport protocol does the MCP Server use?
The MCP Server communicates via SSE (Server‑Sent Events) at the endpoint http://mcp_server:8000/sse.