In this session, we will explore mcp-nomad, a server written in Go that implements the Model Context Protocol (MCP) to interact with a Nomad cluster. This project enables dynamic access to jobs, nodes, logs, and system metrics through a unified and extensible interface. The goal is to demonstrate how artificial intelligence can be connected to real-world infrastructure, providing functionalities such as:

Listing Nomad jobs and nodes

Retrieving allocation logs

Managing Access Control Lists

Collecting real-time system metrics

During the presentation, we will discuss the system architecture, the challenges encountered, and how MCP works—its core principles and how it facilitates communication between systems.

A live demo will showcase mcp-nomad integrated into a real-world scenario. The session will conclude with reflections on possible future developments and integrations, including with other HashiCorp tools.