Check out this MCP here, and contact me if you struggle to get it to work: https://github.com/janspoerer/mcp_obd_ii.

Why I Built MCP OBD-II

My Mercedes-Benz automatic transmission fluid readings weren’t available on my standard OBD-II scanner. Rather than buying a Mercedes-specific OBD-II scanner, I thought to myself that that it cannot be that hard. Why are these things so expensive and why are they all manufacturer-specific?

As I dove into the OBD-II standard, I realized that indeed this standard is very simple and I still do not understand why these scanners are so expensive.

Instead of investing in multiple proprietary scanners for different manufacturers, there are multiple open-source solutions out there already, such as pyobd.

Building on pyobd, I made an MCP server for OBD-II. It is a versatile, AI-accessible interface to vehicle diagnostics.

With this, one can read arbitrary standard or manufacturer-specific information from one’s car.

Why MCP?

The Model Context Protocol makes vehicle diagnostics accessible to AI assistants like Claude. Instead of navigating proprietary software interfaces, you can simply ask questions in natural language:

• “What’s my engine temperature?”
• “Check my fuel system status”
• “List all supported sensors”
• “Are there any diagnostic trouble codes?”

This approach is much more intuitive than having to deal with an user interface (in my opinion).

Getting Started

The library is written in Python and uses the FastMCP framework. It’s designed to work seamlessly with AI agents like Claude Code, making vehicle diagnostics as simple as having a conversation.

Check out the repository for installation instructions and detailed documentation.