The MDF MCP Server analyzes ASAM MDF files (versions 2.x, 3.x, and 4.x) and exposes a rich set of tools for AI assistants to inspect, analyze, and export data. It supports session management, channel discovery, statistical analysis, signal plotting, and data exports to formats like CSV, HDF5, Parquet, and MATLAB. Integrated with UV-based tooling, you can connect your IDE or environment to the server and issue commands to open MDF files, list channels, compute statistics, and generate visualizations or data frames for downstream processing. The server is designed to work well with large MDF files, offering efficient access patterns and an API-friendly interface for LLM-assisted data exploration.

Prerequisites:

• Python 3.8+ and virtual environment support
• Git
• Optional: uv (via install script) for running the server with UV-based tooling

Installation steps:

1. Clone the repository: git clone https://github.com/Shanko-26/mdfmcp cd mdfmcp

2. Create and activate a virtual environment: python -m venv venv

   macOS/Linux

   source venv/bin/activate

   Windows

   venv\Scripts\activate

3. Install in development mode: pip install -e .

4. Run the server locally (example): mcp-server-mdf

Optional: If you prefer using uvx via the recommended workflow, install uv and use: curl -LsSf https://astral.sh/uv/install.sh | sh uvx mcp-server-mdf@latest

Tips:

• The server requires direct file system access to MDF files. Configure your IDE or environment with the correct local paths.
• Memory considerations: for very large MDF files, consider using memory="low" when opening files and limit concurrent sessions.
• If you encounter 'command not found' for uvx, locate its path (which uvx or where uvx) and reference the full path in your MCP configuration.
• Channel names are case-sensitive; use exact names or regex patterns for flexible searching.
• Docker-based deployment is supported as an alternative; ensure Docker is running and adjust volume mounts to expose your MDF data.
• To upgrade to the latest server version in IDEs, use mcp-server-mdf@latest and check version with uvx mcp-server-mdf@latest --version