Experiences with Model Context Protocol Servers for Science and High Performance Computing

TL;DR

This paper discusses the use of the Model Context Protocol (MCP) to unify access to diverse scientific computing services, enabling more seamless and discoverable agent-driven workflows in high-performance computing environments.

Contribution

It demonstrates a practical implementation of MCP servers over existing services, facilitating interoperability and discoverability for scientific workflows.

Findings

MCP enables unified access to heterogeneous research services.

Implementation over mature services shows practical feasibility.

Case studies illustrate improved workflow integration.

Abstract

Large language model (LLM)-powered agents are increasingly used to plan and execute scientific workflows, yet most research cyberinfrastructure (CI) exposes heterogeneous APIs and implements security models that present barriers for use by agents. We report on our experience using the Model Context Protocol (MCP) as a unifying interface that makes research capabilities discoverable, invokable, and composable. Our approach is pragmatic: we implement thin MCP servers over mature services, including Globus Transfer, Compute, and Search; status APIs exposed by computing facilities; Octopus event fabric; and domain-specific tools such as Garden and Galaxy. We use case studies in computational chemistry, bioinformatics, quantum chemistry, and filesystem monitoring to illustrate how this MCP-oriented architecture can be used in practice. We distill lessons learned and outline open challenges in evaluation and trust for agent-led science.