I’m using opencode to build HA Addons for my HomeAssistant on a boat. Works like a charm. But I’m not running opencode on the HomeAssistant directly, I run it on my Labtop in a container and give it ssh access to HomeAssistant. Sure, it can brick your installation, sure it can be a security nightmare and it’s as powerful as running it on HomeAssistant directly.
I was thinking about doing exactly this and googling if it already exists, brought me here.
I think this can be a great tool you did and it’s exactly each users responsibility to use it or not use it. Just don’t come back crying, when it bricks your installation! There are other Addons out there that can do damage to your installation in many ways. If you want safe, use HomeKit or whatever other golden cage.
In general, AI is very capable, when used correctly and guided. In my case, I know what I want it to do and what the result should look like and that helps to write specific prompts that lead to clean results.
I think the fact that it can get things wrong when running unchecked by people who have no idea what they are doing, does not justify calling out people for using AI and ban it or condemning all code written by AI.