Is anyone using Openclaw and tried to use it to manage OpenHAB config? There’s a small thread here but not much other info.
I’ve created a cli wrapper for the REST API that is useful for agents like OC to use, it’s here.
I’ve gotten Openclaw configured now with an agent that has a checkout of my OpenHAB config, the OH docs and openhab-js docs. As it’s on the same network as my OH instance, I’ve given it a restricted login to SSH across to the server, pull the changes down via Git and watch the logs for issues.
Seems to work nicely. I can now tell it issues or problems with rules OR new ideas and it’ll go and fix and PR for my review and/or apply directly. If there’s interest, I might blog about it and share the setup. Anyone else doing this?
Hi,
I have been using my local language models with LM Studio and activated the MCP connection described here.
This works nicely - depending in the size and “intelligence” of the local LLM.
Openclaw so far sits on an isolated computer as I did not yet dare to connect it to my production network.
You might try to give your openClaw a connection via MCP to open HAB which should boost its capabilities.
Would be interested to learn more from your experiments.
Yeah am doing that too and it works well. The only real difference here is you can relatively safely make config changes in a pretty hands off manner.
I’ve written about it here and shared the skill I’m using here in case others would like to try it out too.
My Openclaw and Openhab run on different VMs, so there’s some mucking about creating a restricted user on your openhab server, but after that it’s pretty neat being able to adjust your openhab configs from a telegram chat (while knowing you can easily roll it back if it goes pear shaped).
