This message was deleted.

There's a few things that are different in Cloud. • You can't select LLM. You will be using claude 3.7 using All Hands AI infrastructure. • You won't be able to mount a local drive. You can still upload files. • Right now you can't use a custom sandbox. Although that's an advanced use case. I think those are it off the top of my head. However, OpenHands Cloud has additional functionality. • You will use claude 3.7 with a very high rate limit since it is using All Hands AI infrastructure. • Ability for API keys. • Ease of use of the Cloud resolver. • More fine grained github repository permissions How can we make the documentation better? Is what's missing for you the differences between OpenHands Cloud vs locally?

ty @better-park-34863! I think having a breakdown of capabilities and a section on why one vs the other could help users make sense out of it and probably drive more business for ya'll. One follow up question, how do I gate access to the cloud console? In other words, how do I invite users?

This is good feedback. We will look into enhancing the docs with a better comparison between open source and cloud. User management is definitely something we are working towards right now for OpenHands cloud!

Cool! So how does it work now @elegant-branch-54423? I’ve added it to our repo but no one else can tag the bot except me. What am I missing here?

You mean you've signed up for Cloud, and then you connected a repository. And now you want others to be able to tag OpenHands on the repo? I believe it's because if you are using the cloud resolver, it uses the personal github user that triggered OpenHands. So users need to be signed up as well to be able to tag it. Curious though, what happens when others tag it? Does anything happen (do they get an error?) or nothing happens?

Nice! Yes it actually works if they also sign up, which is pretty easy since I already authed the gh app. I think there would be a lot of value in having a concept of a team with access control, analytics dashboard and more than anything a way to share threads.

For sure! Pretty sure these are on the eventual roadmap! Appreciate the feedback.

Love these ideas Sebastian! Curious, are there specific analytics/metrics you're looking to track with OpenHands? We have some ideas here but wondering if there are 2-3 that come top of mind that are important for you.

General usage for sure, like threads/executions per user and the incurred costs. I guess the agent will show up under the GH stats page anyway, so we could find out what it's doing there.