Colabfit Mcp

Colabfit Mcp is a Model Context Protocol (MCP) server designed to connect conversational AI agents (such as Claude, OpenAI agents, and Gemini) with local scientific data discovery and machine learning workflows. It enables AI agents to search and download scientific datasets from the ColabFit repository, train MACE (atomistic) interatomic potentials locally using KLIFF and KLAY, run simulations/energy calculations, validate models with OpenKIM test drivers, and manage datasets and models. Ideal for researchers and developers in computational materials science, it standardizes access and orchestration via the MCP protocol, leveraging both CPU and GPU hardware.