If you manage inf
 rastructure, you already know two things: documentation is always one step behind reality, and uploading your network topology, VLANs, and internal IPs to a cloud AI is simply not an option.

That’s why I’ve integrated a Local AI Agent directly into RackDoc, my open-source, local-first tool for documenting servers, racks, and technical dependencies.

No Cloud, No Telemetry, and Air-Gapped Ready

The agent runs entirely on your own infrastructure, connecting directly to your local instance's inventory:

  • Total Privacy: Your IP ranges, hostnames, and network diagrams never leave your machine or data center.
  • Available When Everything Else Is Down: Because it doesn't rely on external APIs, it works in isolated networks or during connectivity outages—precisely when you need to check your inventory the most.
  • Zero Proprietary Dependencies: No token fees, rate limits, or SaaS subscriptions.

Real-World Day-to-Day Use Cases

Instead of wasting time cross-referencing port tables or scattered files during maintenance or an outage, you can now query your documentation directly:

  1. Dependencies & Single Points of Failure: "Which services and VLANs depend on Switch 02 in Rack B?"
  2. Physical & Network Capacity: "How many free U slots and open ports are left in the main cabinet?"
  3. Quick Audits: Spot duplicate IPs, undocumented hosts, or inventory mismatches in seconds.

Code & Deployment

RackDoc remains lightweight, self-hosted, and straight to the point. The source code is live on GitHub if you want to spin it up in your homelab or internal environment and take the agent for a spin:

👉 github.com/mario-ezquerro/rackdoc

If there’s a specific query or integration you’d find useful in your day-to-day sysadmin workflow, feel free to open an issue or drop a comment below.