Introduction

• Matthew Feickert (University of Wisconsin (US))

Vibe Plotting - Using LLM's to make physics analysis plots

• Gordon Watts (University of Washington (US))

Building the HEP Agent Stack, MCP Servers, CLI Tools, and Skill-Based Workflows for Scientific Analysis

• Mohamed Elashri (University of Cincinnati)

This talk presents a practical agentic AI stack for high-energy physics through three projects: `root-mcp`, `cerngitlab-mcp`, and `inspirehep-mcp`. These tools expose ROOT data analysis, CERN GitLab software discovery, and INSPIRE-HEP literature workflows through MCP servers, command-line tools, and lightweight `SKILL.md` files. The talk focuses on the shared-backend design behind these tools, the tradeoffs between MCP and token-efficient SKILL + CLI workflows, and how this approach can support reusable, reproducible HEP research infrastructure.

MITRA: An AI assistant for Knowledge Retrieval in Physics collaborations

• Abhishikth Mallampalli (University of Wisconsin Madison (US))

We present MITRA, a Retrieval-Augmented Generation (RAG) system developed to assist particle physicists in navigating collaboration documentation. The system facilitates the extraction of relevant information from analyses and shift documentation and provides answers to user queries with direct citations to the original sources. While the current implementation uses CMS analysis documentation, the underlying workflow is designed to be experiment-agnostic. The pipeline: including document ingestion, embedding generation, and retrieval, is modular, allowing for adaptation to the documentation of other collaborations. To meet the privacy requirements of HEP experiments, MITRA operates entirely on local collaboration infrastructure using self-hosted, language models. This ensures that internal documents such as unpublished results remain within approved servers and are not exposed to external APIs. This also reduces cost, allowing the system to scale across the user size of HEP collaborations and across time. We describe the motivation for building MITRA, performance of the retrieval pipeline, and outline ongoing work toward more complex, multi-step reasoning workflows. https://arxiv.org/abs/2603.09800