Re:Jellyfin 12.0
by joshstrange on Tuesday September 08, 2026 @07:13AM
(#49608695)
I’ve set up literally hundreds of different self-hosted applications over the years on my homelab. I’ve configured the reverse proxy by hand, setup the docker containers, etc.
Now I can just tell Claude “Setup XYZ app on my server” and it does it all and leaves a note in my Obsidian vault documenting the service (setup instructions, backups, hostname, etc).
Trying out new things has never been easier now that the effort is near-0 to kick the tires.
I would caution against this approach. It's fine to have Claude write a setup script, or whatever you need, but don't run coding agents as your main user on your production machine with full access to your files or whatever your "homelab" consists of.
Re:Jellyfin 12.0
by bigstrat2003 on Tuesday September 08, 2026 @03:00PM
(#49615088)
Yep. You should never let an LLM have access to the system. It's irresponsible as all get out.
Re:Jellyfin 12.0
by threecheese on Tuesday September 08, 2026 @08:22PM
(#49619075)
Your personal computer? Eh. This afternoon Claude code diagnosed an issue I’ve had with my 4K displays since Tahoe, it would have taken me a week just to learn enough to try to figure it out. Now I have a cli tool which fixes it. I get a weekly report of Chezmoi drift across all my machines, including what I changed. This took me 30 minutes to set up. There are so many things it can simplify.
A production server with user data on it? I wouldn’t dare. My backed up MacBook? Truly valuable.
Next step: set up a scheduled job to read the logs to find any issues before they bite.
Being realistic, most of our home lab setups wouldn’t come close to “production ready” even without AI generated bugs. Relatively stable because they’re largely un-stressed, yes. Good enough for personal use because you can log in and fix anything, sure. Full of bugs, footguns and brain dead decisions because this is our hobby and not our 40 hour a week paid job? Also yes. Our new AI overlords will simply increase the number of buggy things we can deploy.
Hmm, I feel like I have a pretty decent … well, not production-ready, but it’s all in Ansible and I did have to rebuild it from that once, which ironed out most of the bugs in that and everything gets built according to the templates I have for Docker and non Docker setups.
Plus I have “itguy” now, a CLI that can deploy everything, check on status and show me logs. And it’s available via Slack and as an Alfred workflow. So yeah, I think that’s pretty close to production ready.
LLMs have been AMAZING for home server maintenance and improvements. I have completed SO many of my "some day" tasks. I never have to read another fucking log file ever again. It has taken all of the friction out of the hobby. It's all* fun now.