Plug-and-Play: Connect PowerMCP and PowerSkills to Your LLM

Qian Zhang – Harvard University

PowerAgent (https://poweragent.seas.harvard.edu/) is an open-source community focused on accelerating agentic AI development for the power systems domain. In its first two months, it has attracted 100+ followers. Our goal is to bridge cutting-edge AI with the real-world needs of system operators, electric utilities, and power system researchers.

This tutorial is organized around a practical question: If every power engineer has a general-purpose LLM on their laptop, what is the fastest way to make it engineering-useful? We argue that the key is plug-and-play tool access plus reusable, maintainable skills that map directly to engineering workflows.

We will present a hands-on demo showing how to build and compose PowerMCPs (tool interfaces) and PowerSkills (workflow-level capabilities). Participants will see how to connect an LLM to domain tools safely, reduce repetitive work, and scale from simple tasks to more complex workflows through skill composition.