I have not yet tried dthat. but thats the next step i should take

Its has produced great documentation for my homelab. Thats what it did best, could not have done it without having it conduct the tasks it was asked to do

[ranjan@llmbox Homelab Wiki]$ tree -L 2
.
├── Clients
│   ├── CachyOS-Laptop.md
│   └── README.md
├── Infrastructure
│   ├── README.md
│   ├── Router.md
│   └── Switch.md
├── README.md
├── Servers
│   ├── dms.md
│   ├── llmbox.md
│   ├── mediabox.md
│   ├── nasbox.md
│   ├── README.md
│   └── Router.md
└── Services
    ├── AdGuard-Home.md
    ├── BentoPDF.md
    ├── Beszel.md
    ├── Caddy.md
    ├── Collabora.md
    ├── Degoog.md
    ├── Dockhand.md
    ├── Flaresolverr.md
    ├── FMD.md
    ├── Food.md
    ├── Forgejo.md
    ├── Glance.md
    ├── Gonic.md
    ├── Homepage.md
    ├── Immich.md
    ├── Invidious.md
    ├── IT-Tools.md
    ├── Jellyfin.md
    ├── Jotty.md
    ├── Linkding.md
    ├── Llama-Swap.md
    ├── Metube.md
    ├── NFS.md
    ├── Ntfy.md
    ├── Omnitools.md
    ├── OpenCloud.md
    ├── Prowlarr.md
    ├── qBittorrent.md
    ├── Rackpeek.md
    ├── Radarr.md
    ├── Radicale.md
    ├── README.md
    ├── Redlib.md
    ├── SABnzbd.md
    ├── SearXNG.md
    ├── Seerr.md
    ├── Signal-DMS.md
    ├── Sonarr.md
    ├── Speedtest.md
    ├── Tailscale.md
    ├── Termix.md
    ├── Transmission.md
    ├── Uptime-Kuma.md
    └── Vaultwarden.md

Look at the my setup…

1. The Intel Xeon E-2224G was a server/workstation processor with 4 cores, launched in May 2019
2. DDR4 32 GB non branded sticks

Its ancient cheap hardware. Whats stopping you?

Does your setup handle automagic updates, how do you handle prompt injection if so? Or, just error correction/logging?

Figure it out yourself with your agent. What suits your use case? what would you like the agent to do? how much of a risk can you take with your agent. It would vary depending on so many factors

I knew dockhand/portainer would do docker updates better, i knew auto updates can be setup via cron for os updates, etc.

i am neither a sys admin, nor a programmer, i just run a hobby homelab and like to tinker and learn. its a good enough usecase for me to explore the possibilities

So dont do it!

Its a learning experience, how can a coding agent be used in a non coding way? is it better or worse? i guess i have my answers now,

this may not be the ideal usecase, but it surely shows that these agents can be used for other things.

Its a homelab, not a commercial production environment, agree with you, but I am not too worried about it.

At 56, i was laid off from a Fortune 500 company, so i hear you. Today I am without a job just trying to learn and keep up everyday.

Edit:Spellings

Welcome! if you have questions on ik build parameters for optimizations feel free to ask, I will try my best to answer

Thanks! Its a fun experiment!!

Its an MoE model (https://en.wikipedia.org/wiki/Mixture_of_experts), only 3B parameters are actually active

I have 32GB RAM

my sudoers file restricts what the llm can actually do, also I have robust backups can can spin up any of my servers really quickly, I am not that worried and just like you deal with human errors, you can deal with agent errors.

so far this has been running for a month, no scares or unexpected behaviour other than looping on a task somethimes

Open-source models provide complete access to the entire model architecture, training methodology, and weights. This comprehensive access includes the model code, architecture design, training scripts, and parameter weights under licenses like MIT or Apache.

Open-weight models represent a more limited approach to model sharing. These models release only the trained parameter weights while restricting access to training methodologies and code. Also under MIT or Apache licenses

An Open-Weight Model is an AI model whose core components are publicly released, allowing anyone to download it. This lets users run the model on their own computers, study how it works, and even modify it for their own specific needs.

Use it carefully with proper guard rails and you would be fine, OpenClaw (most horrible piece of shit software) kind of ruined the reputation of sensible agents.

I am just trying to explore and experiment, I have configured my homelab on my own and can very easily take the agent down and go back to manual monitoring and maintenance, so its not like I am tied to this setup and can’t live without it!

The model is an open weight model, Google or Open AI did not create it. I use my electricity at home, So no it doesn’t

Fun fact: you don’t have to, I expected to be voted down on this post, but I have had fun setting it up and wanted to share

Yes, the download link for the model is in the original post

Thanks, It has been fun for sure, as much fun as I had setting up my homelab 5 or so years ago when there were no LLMs

The local LLM is run on the homelab, just like immich is run on your homelab and doesnt talk to google photos is any way, its the same for my model, self contained, inhouse with no data leaving my network

Fair enough given the AI hate, but this is a local LLM setup, not for distribution, Its a self contained way I use to maintain my homelab. Some may find it useful, some may not.

Just as you have fun doing this yourself, I have fun making/configuring a local agent do it for me