please please give us an option to make agent always accept and run all commands without waiting for user confirmation.
its pretty tiresome to be multitasking, thinking the agent has finished but turns out to not even be close to finishing because it was waiting for your confirmation to run a terminal command
i get the risk perspective, but give option to users that do understand the risks to toggle it on/off
I second this!
+1.
I think there needs to be some improvements before phasing this feature in.
Eg. when I open two projects (backend & frontend) their parent folder, sometimes commands run in the parent folder and creating files there without noticing.
+1 will be very helpful in a bugfix with running CI
This should work when you’re using Composer with Agent mode and YOLO enabled.
That said, I find it sometimes reverts to asking your approval on every command despite having this configuration.
When this happens, I tell the model:
“Remember, you’re in Agent mode with YOLO enabled, so you can run terminal commands yourself without my help.”
This usually fixes the issue and the model returns to running commands on its own.