Autonomous Network Compliance with MCP and AI Agents
This project takes the shared-lab pattern into IP networking operations. Participants build three structured agents — config inventory, drift / compliance, topology diagram — and then a fourth completely freehand agent of their own design. The freehand lab is the hackathon-vibe finisher the customer asked for.
.svg){kind=small}
About The Lab
Prerequisites
Audiences
Lab Architecture
The Lab architecture integrates NetBox as the source of truth, an MCP NetBox Server as the secure API bridge, and Claude Code as the AI runtime to create a comprehensive environment for AI-driven IP network automation, which includes four specialized agents — Config Inventory, Drift & Compliance, Topology & Diagram, and Custom AI Agent — that allow the user to test end-to-end multi-vendor (Cisco NX-OS, Juniper JunOS) use cases, from extracting normalized configuration inventory and auditing compliance drift to generating Mermaid topology diagrams and running extensible custom workflows, producing structured outputs (inventory.json, drift-report.json, topology.mmd, custom-output.json) for understanding, visualizing, and automating the network.
Why this Lab ?
This project takes the shared-lab pattern into IP networking operations. Participants build three structured agents -- config inventory, drift/compliance, topology diagram -- and then a fourth completely freehand agent of their own design. The freehand lab is the hackathon-vibe finisher the customer asked for.
Lab Objectives
- Build three cooperating IP networking agents following structured guidance, then a fourth freehand agent.
- Use real model routing for config parsing, drift detection, and diagram generation.
- Combine static configuration artifacts with live mock MCP data covering inventory, ownership, and tickets.
- Practice human approval before any agent write action.
- Finish the project with one freehand networking agent the learner would actually want in their day job.
