MCP-CAN: Virtual CAN + MCP Server

An MCP server purpose-built to surface vehicle CAN/OBD data to an LLM/SLM. It simulates ECUs on a virtual CAN bus, decodes via a DBC, and exposes MCP tools over SSE—no hardware required by default.

Highlights

• MCP server for CAN/OBD → LLM/SLM (tools + DBC metadata over SSE).

• Virtual CAN backend (python-can) out of the box; optional SocketCAN/vCAN on Linux.

• DBC-driven encoding/decoding via cantools.

• ECU simulator that streams multiple messages plus demo OBD-II responses.

• MCP server (SSE) exposing tools for frames, filtering, monitoring, and DBC info.

• Typer CLI: mcp-can (simulate, server, frames, decode, monitor, obd-request).

• Dockerfile + docker compose for server + simulator.

• Unit tests, type hints, lint config (ruff, mypy).

Repository Layout

• src/mcp_can/

  • cli.py – Typer commands

  • bus.py – python-can helpers

  • dbc.py – DBC loading/decoding

  • config.py – env settings (MCP_CAN_*)

  • models.py – simple dataclasses

  • simulator/runner.py – ECU simulator + OBD responder

  • server/fastmcp_server.py – MCP tools (SSE)

  • obd.py – minimal OBD-II request/response helpers

• vehicle.dbc – sample CAN database

• simulate-ecus.py, can-mcp.py – entrypoints

• docker/compose.yml, Dockerfile

• tests/ – unit tests

Prerequisites

• Python 3.10+

• (Optional) Docker / Docker Compose

• (Optional) Ollama if you want a local LLM backend

Install (Python)

From repo root:

Quickstart (Simulator + MCP Server)

Two terminals:

Single-process (helps on Windows if virtual backend doesn’t share across processes):

Sample interactions:

MCP Inspector (GUI for your tools)

Use the official Inspector to explore and call your MCP tools without writing a host:

When prompted, connect to your server:

• URL: http://localhost:6278/sse

You can then:

• List tools and resources (read_can_frames, decode_can_frame, filter_frames, monitor_signal, dbc_info).

• Call a tool (e.g., monitor ENGINE_SPEED for 5 seconds) and view JSON output live.

Using with Ollama (local LLM)

1. Ensure Ollama is running: ollama serve and pull a model: ollama pull llama3

2. Run simulator + MCP server (see Quickstart).

3. Point your MCP-capable host at http://localhost:6278/sse and configure its model endpoint to http://localhost:11434 with your model name (e.g., llama3).

4. Prompt the host: “Monitor ENGINE_SPEED for 5 seconds” or “List all DBC messages.”

If you need a minimal host, pair @modelcontextprotocol/sdk with Ollama (see SDK docs) or use Inspector for manual tool calls.

Example host config (OpenAI-compatible endpoint to local Ollama):

CLI Reference

• mcp-can simulate – start ECU simulator using vehicle.dbc.

• mcp-can server [--port 6278] – run MCP SSE server.

• mcp-can frames --seconds 1.0 – capture raw frames as JSON.

• mcp-can decode --id <hex|int> --data <bytes> – decode a single frame.

• mcp-can monitor --signal <NAME> --seconds 2.0 – watch one signal.

• mcp-can obd-request --service <hex|int> [--pid <hex|int>] – demo OBD-II request.

Configuration

Env vars (prefix MCP_CAN_):

• CAN_INTERFACE (default virtual)

• CAN_CHANNEL (default bus0)

• DBC_PATH (default vehicle.dbc)

• MCP_PORT (default 6278)

You can set these in a .env file at repo root.

Docker

Build:

Run (combined server + simulator):

Compose (from docker/):

Development & Testing

Troubleshooting

• No frames? Ensure both simulator and server use the same interface/channel (virtual/bus0 by default).

• DBC missing? Set MCP_CAN_DBC_PATH or place vehicle.dbc in repo root.

• Docker networking: expose 6278 so your MCP host can reach SSE.

License

MIT (see LICENSE). Educational/prototyping use only—use certified hardware for real automotive work.