AMI is an MCP server designed for ATLAS physicists that wraps the ATLAS Metadata Interface (AMI) and the PMG cross-section database, making them accessible as powerful tools for Large Language Models (LLMs) like Claude. This integration allows LLMs to directly query ATLAS metadata, enabling the discovery of Monte Carlo (MC) datasets by various criteria, retrieval of crucial physics parameters such as cross-sections, filter efficiencies, and k-factors, and comprehensive validation of PMG hashtag classifications. It streamlines the process of looking up dataset provenance and executing arbitrary AMI command strings, significantly enhancing data exploration and analysis for particle physics research.
