Whats the state of play with letting an LLM have access to the whole Home Assistant system? are there any integrations out there that allow this?
My goal if for AI to see my system as a whole, so I can request changes to automations etc on the live system, as opposed to copying an automation, pasting it in to an LLM, then pasting the results back to my system.
I have to reply to use this emoji:
Why? is it a security concern, or is there genuinely a simple way to do this that I am missing?
I have briefly given this a whirl, but the suggestions were often not suitable:
I just wouldn’t put an LLM in charge of a computer.
I’m not saying it will be in charge as such, I just want it to be able to see my system as a whole, so if I ask it to make some changes to an automation, it is aware of other automations that the change could conflict with.
I get that it’s not always great and without checking the code it can make some serious ****ups, but it has saved me hours building automations - and has enabled me to build custom cards that I would have no chance creating from scratch.
It would just be handy to have it more integrated somehow.
At your own risk… 
Just found this which is almost exactly what I was looking for!:
Didn’t seem to work for me. There was no integration to add after installing.
Haven’t looked at it for a while - but isn’t it an app (add-on)? The developer was talking about turning it into an integration, but I don’t know how far he got with it.
Edit: There’s a page discussing it here:
I’ve got the MCP server running and now everything’s a freaking dream. Want a complex automation configured? Just describe it and…. done. Something weird happening? Ask about it…. agent investigates, proposes solution, gets permission to fix… done.
THIS is how I’ve always dreamt things should work!
I am curious. Can you show an example?