Hey folks! If you're using the OpenHands CLI in y...
# feedback
q
Hey folks! If you're using the OpenHands CLI in your day to day I would appreciate if you used the experimental UI (add
--exp
flag to your desired way of running the CLI) We plan on making this the default experience and would be great to hear feedback before its done
❤️ 6
👀 1
👍 1
f
One tiny thing I notice is the experimental CLI doesn't wrap the text into new lines as user types, which makes it a bit hard to see what was typed earlier. OTOH the current CLI does wrap the text I think
acknowledged 1
thanks on a stick 1
c
I had a silly surprise when I first tried it: I entered something, and it started humming, and nothing, and nothing… Then I saw the confirmation box on top right 😅 I mean, I’m a bit dense, I didn’t expect that. My eyes were on bottom left, and it was on top right 😂 Is this the right UX? I feel maybe it’s PEKBAC, but was there a reason for it to not be where, I think, the user of a CLI expects it? 😂
q
Ahhh that's true! cc @breezy-kilobyte-39430 for help with the designs (specifically for confirmation mode in this instance) I def took some creative liberty for the sake of getting this out quickly 😅
s
I've seen some cases where the confirmation box on right side has a lot of text and the "buttons" get shifted offscreen. It looks just stuck in this state but if I realize its waiting for input and hit enter then it accepts and continues.
s
It would be helpful if there was some sort of a log somewhere, when I use the same settings as the webUI I get no errors just Working... and nothing happening. When I play with the settings I can some errors like "litellm.InternalServerError: InternalServerError: OpenAIException"
which makes me think the settings that work for the webUI (which I know runs docker containers) are the correct ones to use, but I still can't get them to call LM studio like the webUI does