OpenML MCP Server

Example Use Cases

• 1

  Fetch metadata and descriptions for OpenML datasets (e.g., dataset features, descriptions, and IDs) directly from a chat or agent interface.

• 2

  Search and list OpenML tasks, flows, and runs with basic filtering to discover benchmarking tasks or experiment results.

• 3

  Integrate OpenML queries into Claude Desktop or other MCP-compatible clients to enable interactive exploration and retrieval of experiment artifacts.

Description

This project wraps major OpenML GET endpoints as MCP tools, enabling interactive queries from MCP clients such as Claude Desktop. It provides endpoints for datasets, tasks, flows, runs, evaluations, setups, and studies, with list and filtering support where the OpenML API allows. The server is built with the mcp Python SDK (FastMCP) and is configured to run via the uv runner; an OPENML_API_KEY environment variable is used to authenticate requests. The main benefit is quick, agent-friendly access to OpenML data without writing custom HTTP code.