#Assist - Switch from local to cloud issue

1 messages · Page 1 of 1 (latest)

split skiff
#

(Running Home Assistant Core) Just switched from using piper and faster-whisper to HA Cloud. Now whenever I try to update any assist settings such as stt or tts i recieve:

Source: components/websocket_api/connection.py:315
integration: Home Assistant WebSocket API (documentation, issues)
First occurred: 12:08:11 PM (1 occurrences)
Last logged: 12:08:11 PM

[281470487012416] Error handling message: extra keys not allowed @ data['prefer_local_intents']. Got False (invalid_format) Chris from 192.168.0.116 (Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/128.0.0.0 Safari/537.36)```
#

im also getting it with other things such as action: assist_satellite.announce target: entity_id: assist_satellite.tars message: "test" for which i receive: Failed to perform the action assist_satellite.announce. extra keys not allowed @ data['sequence'][0]['target']['message']. Got 'test'

mental forge
#

What version of HA are you running?

split skiff
#

@mental forge 2024.11.3

ionic dune
#

Upgrade.

mental forge
hardy zealot
#

Good evening. I am doing as much pre-work as I can while waitng for my VPE to arrive. Does the fall back to LLM only work with a voice assistan? I have tried general knowledge prompts from the HA front end Assist and It just gives me an error that it can't find a device with the name of the prompt. How old is X (for example). I am running 12.4 and the radio button for fall back is on. No errors in the log. It just never attempts to fail over in the front end UI. Does it not work that way in the chat UI? Thanks!

surreal shuttle
#

So usually you make an llm conversation agent, set that as your voice pipelines agent, and set prefer local so it goes to ha first, then the llm if no intent is matched

#

You also have to choose whether or not the llm conversation agent is allowed to control and see devices and such

#

But also please make a new thread instead of hijacking this one 😉