Using cloud agents with environment.json

Support Help
1

Hi,

I’ve set up an environment using .cursor/environment.json. I have a custom Dockerfile (to install docker-in-docker), install/update and start scripts.

Those all seem to be working, and if I create a new agent from my repo/branch, it’s able to work.

How can I have Cursor recognize the environment has now been set up, to turn off the “Set Up Environment” flag in the prompt window, and the popups on the Desktop/Terminal tabs that warn “Environment not configured”?

I also have a question about multi-root workspaces. I saw there’s a field repositoryDependencies in environment.json; what would an example look like? I tried listing some repositories there, but it seems to have ignored them. Where would they get checked out to, if it did work?

5

The confusing part is that the environment can be working technically while the UI still treats it as not configured.

For cloud agents, that status needs to reflect the actual setup state pretty closely. Otherwise users end up debugging the wrong layer: Dockerfile, install script, repo checkout, or just a stale environment flag.

7

Once I merged the PR with the environment config into our default branch, the “Set Up Environment” hints went away - looks like that’s all that’s needed, it just needs to be on the default branch.

10

Hi @mzs Thank you for following up and sharing your findings on this.