I suspect cloud agent setup will be revamped. That said I’m looking to use it now. It’s really cool how the interactive setup works but at the same time I feel it reinvents the wheel for setting up a vm with questionable gains. At the end of it I wasn’t able to take a snapshot that worked with an init script 
Docker supports secrets in builds by mapping env from the host. Is something similar available for cursor vm setup? I don’t want to just upload my .env.local and make that available to ai agents
In general it’s confusing when secrets added to cursor agents are available — it seems like not during vm setup, which is where I need them (build secrets). I see they’re available at runtime but I don’t need build secrets there. Less important but still — what happens when a team and a user have a secret with the same name?
I didn’t find it clear how to manage general config (non secrets).
Are there any self hosting options or more integrations and config planned here? Docker compose would be great.
Machine sizes? Are your base images available for extension?
Why is there only one global config for all this? Shouldn’t we be able to have a setup per agent type?