The agents sometimes might not produce the required functionalities half way in the process, making the coding flow not so fluent.
An “await” button can stop the current response process and wait for the user’s command, which might be a correction to the development steps, or letting the model notice mistakes or ignorance that it has made.
It would be nice to be able to restore check points and reject changes when awaiting and start from the middle of an agent’s develop process.
3 Likes
Like being able to correct the model’s mistakes mid-response without interrupting the flow?
yeah, like giving it a correction prompt so that the flow goes towards the right direction, add additional requests for the model to develop, or let it notice mistakes that it has made
Isn’t this the same as hitting stop and then sending a new message to the agent?
right, I see that there are small add in “+” buttons in the flow after the agent stops, which can be used to resume to the place where the code went wrong
