First off. I love this feature, and it will probably be very useful for us. In particular for scripting and CI/CD pipelines.
A couple of concerns I have are:
The output is not very verbose in non-interactive mode. I would like to see more of what’s going on (similar to what the interactive mode shows). This is especially true when a step hits a pause. It would be nice to know what is actually being “thought” or ran.
Please have some sort of timeout and output for when a command get’s stuck (where it would usually be asking for permission to run a command). While running with the -f flag helps avoid this, it tripped me up multiple times with no idea why, when it was trying to run git commands.
It would be really nice to have a way to connect to MCPs in the CLI (including non-interactive) … which would allow us to really supercharge CI/CD workflows.
That’s what I have for now, but I am sure I will have more thoughts and ideas as time goes by. Again, thanks for the great feature. Looking forward to seeing how it evolves.
Running text output. It was good in some follow-ups when it was reporting:
”Ran Search”
”Read ….”
But yeah, I think it would be great to have it say when it’s attempting to run a tool command, and would have required a user interaction. In my testing, I basically just had to kill the command after waiting helplessly (and blindly) for a few minutes.
Also, some train of thought hints would be really helpful for those of us who are used to using the tool in an IDE environment, where we see that kind of output.
