One frustration I keep hitting: every transcript labels me as "Speaker 1" (or 2, or 3) instead of recognising me as the owner. I'd happily do a voice enrollment or onboarding step so it knows who I am. I'm fairly sure I already did one, but it doesn't seem to remember me across recordings.

Has anyone got this working reliably?

• Is there a voiceprint or speaker profile setting I've missed?
• Does it only apply to new recordings, or can it relabel old ones?
• Does it work differently on the app versus the web?

Any tips appreciated, because renaming speakers on every recording is getting old.

I've been a long time hermes user, so was excited to learn that rabbit supported it. I got a second hand device off Ebay and set it up (I'm in UK, these aren't sold locally).

This thing gets stuck all the time and leaks ACP sessions! I have created a watchdog hermes skill to clean up leaked sessions and restart rabbit-agent to get it working again.

There is also other minor things - like when Hermes needs some permissions, the prompt on the device says "Claude is requesting permission" - they haven't done some basic QA testing.

Is there a place where you can file bugs?