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

2025-08-25

AI Summary

To address the critical challenge of adapting large language model (LLM) agents to heterogeneous, security-constrained computational infrastructure (CI) APIs in scientific workflows, this paper proposes a unified interface architecture based on the Model Context Protocol (MCP). We design a lightweight MCP server that encapsulates diverse research services—including Globus (Transfer/Compute/Search), Octopus, Galaxy, and Garden—as discoverable, invocable, and composable context modules, enabling cross-domain, secure, and scalable agent-driven scientific support. The architecture comprehensively covers core scenarios: data transfer, computation execution, and state monitoring, and is empirically validated across computational chemistry, bioinformatics, quantum chemistry, and filesystem monitoring. Key contributions include: (1) the first systematic application of MCP to CI integration; (2) a substantial reduction in the barrier for LLM agents to access scientific infrastructure; and (3) identification of fundamental challenges in evaluation and trust within agent-based science.

Technology Category

• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search

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.

Problem

• Integrating heterogeneous research APIs for LLM agents
• Providing unified interface for scientific workflows
• Enabling secure composable research capabilities

Innovation

• Model Context Protocol unifies research APIs
• Thin MCP servers integrate mature scientific services
• MCP architecture enables discoverable composable workflows