MCP at the LHC: Coordinating the World’s Largest Machine
summary
Prototype MCP server connects MS Project via the scheduling backend to an on‑prem LLM and returns conflict analyses rather than raw tables, shaving context by pruning fields and moving constraint logic into code. The demo calls backend APIs, analyzes thousands of tasks, and surfaces four conflicting tasks for LSS L2 on 2 Feb 2026. Implementation uses a Python MCP endpoint + CERN‑hosted LLM and proves workable while showing the key rule: logic in code, not in the model.
AI-written summary · about
chapters (11)
- 0:00MCP at the LHC intro
- 0:23LHC scheduling basics
- 1:48LHC maintenance constraints
- 3:59MS Project challenges
- 4:43Model Context Protocol
- 5:46MCP use case in LHC
- 6:28MCP server implementation
- 8:13Shifting logic to MCP
- 9:39MCP client demo
- 10:20Key takeaways
- 11:04Q&A
related talks
02h 49mMCP Servers: Best Practices, Design Choices and Their Consequences
32mMCP for dummies: Feed more context to LLM
46mMCP for LLM Agents, Infrastructure APIs & Data Services
25mFrom Idea to Integration: An Intro to the Model Context Protocol (MCP)
32m30 minutes to understand MCP (Model Context Protocol)
47mMCP Demystified: Cutting Through the Hype-What It Is, What Isn't & Its Role in the Agentic RAG Stack
35mMCP in Action: Real-World Use Cases & Design Patterns
43mLLMs, Meet Your Tools: The Magic of Model Context Protocol