HN Explorer

Ask HN: AI for Home Lab Infra?

1 points by voakbasda ↗ HN
In these days of AI agents capable of breaking into remote systems, are there any up for the task of maintaining a home lab with public infrastructure (e.g. DNS, SMTP, etc.)?

I have dozens of Linux machines scattered throughout my property that need regular updates, migrations and general maintenance. To start, I imagine cloning one into an VM and letting an agent give it a go. If the results looked good, I would give them a crack at the live system on the edge. Ultimately, I would love to have an agent handle all of my infrastructure.

That said, I have been maintaining my own infrastructure for almost 30 years, so the idea of handing it off to an AI seems far fetched. Heck, it would take a senior engineer to untangle the mess that I’ve cobbled together over the years. Yet I cannot deny the impressive results that I am getting from my current casual level of AI chat.

So what are people using for this task? How well do they do the job?

6 comments

[ 0.39 ms ] story [ 3.2 ms ] thread
andsoitis
> I have dozens of Linux machines scattered throughout my property that need regular updates, migrations and general maintenance.

One idea is to move from "dozens of Linux machines scattered throughout my property" to just one.

gitowiec
I think the most important work will be preparing the context for an agent (eg OpenClaw). Then add some skills (eg use ssh to visit specific machine). And finally the prompt. Maybe you should check what agent has the best tools for DevOps?
frangonf
I would start by using AI to port the tangled mess into as much a tidy IaC as possible, with git backed declarative config tools, to prepare the safe ground where agents have context, can get loose and easily review, lint/check or test in throwaway VM before applying it.

A fleet of Openclaws (or equivalent) with control of machines for 100% hands off management seems like a russian roulette to me as of today, but a 90% handoff to AI for reviewing current config of X in our system, reading docs and configuring a deterministic tool is achievable and somewhat safe as you can easily review and trace what has been done.