When running an agent in the cloud, I have it setup so it can run my repo in the VM, then use the browser to test changes. That’s all working well when running an agent manually. However, when setting up an automation, the agent it spawns doesn’t seem to have access to the same environment or tools to also use the browser to demo and test changes. When creating the automation, I have the “Use Configured Environment” enabled as well. When asking a follow up question after an automation triggered (in the cursor.com UI), saying something like “Test that change in the browser”, it says it doesn’t have access to any MCPs to do that.
Should these automations be able to test the changes as they normally would when running the agents manually?
