MCP x Quantum Science Hackathon

MCP x Quantum Science Hackathon

To empower researchers to integrate AI agents into their own scientific workflows and experience the benefits of AI-assisted research, the Stanford Q-FARM and the Path Integral Institute are co-hosting the “MCP × Quantum Science” Hackathon. This event aims to make the Model Context Protocol (MCP) accessible to the broader research community, enabling participants to connect AI agents directly to their tools, data, and research tasks.
Over the course of two focused days, the goal is to create dozens of practical applications where AI agents assist with real scientific problems — not just information gathering, but actively contributing to computation, simulation, analysis, and idea exploration.
Unlike a typical hackathon, this event is designed specifically for researchers — including students, postdocs, and faculty — who:

• Are curious about working with AI collaborators in their research
• Use Python and have research tasks involving Python-based tools

The hackathon will begin with an overview and a step-by-step guide on building MCP servers. It will conclude with group demos, where participants will showcase the AI agent powered by the newly built tools.

We believe this event will significantly accelerate the integration of AI into scientific research at Q-FARM, Stanford, and beyond — helping researchers turn AI agents into powerful collaborators for discovery.

Background

The rapid advancement of large language models (LLMs) over the past two years has enabled AI to perform complex tasks such as coding and solving advanced mathematical problems. However, AI has yet to see widespread application in scientific research beyond a few isolated examples.

Can we envision an AI collaborator that not only gathers information but also actively participates in research — running experiments, executing computational tasks, and exploring new scientific possibilities? We believe the answer is yes.

The Model Context Protocol (MCP) is an open-source protocol developed by Anthropic that acts as a universal adapter, connecting AI agents to a wide range of data and tools. Much like a USB port for research infrastructure, MCP allows you to convert your tools into MCP servers, making them accessible to any AI agent compatible with the protocol. With natural language instructions, you can teach AI to operate your code.

For instance, if you have three Python functions that construct a Hamiltonian, diagonalize it, and plot the eigenvalues, you can turn them into an MCP server with three tools — enabling the AI to perform these tasks through simple language commands.

Thousands of MCP servers already exist, though few are currently used in scientific contexts. By converting research tools into MCP servers and using AI agents as “co-pilots” in scientific research, we unlock powerful new possibilities, including:

• Dramatically increasing research efficiency
• Lowering the barrier to learning and using new scientific tools
• Enhancing collaboration across disciplines

This approach could fundamentally transform how research is conducted.