Yoooo,
Maybe we could do some file management system, or just cleverly use git to have an option to run and debug multiple agent outcomes in an asynchronous manner?
The goal is to test it in isolation and omit conflicts.
Imagine I have 3 tasks that could be done separately from the MAIN branch.
I would love to type 3 prompts for each task, preview the outcome, see the console, write a few update prompts for each problem, and deliver 3 of them in a time of 1.2 tasks.
We go cloud now. But it is missing all your debugging setup there, preview, and you need multiple messages to get what you want instead of a single shot.
Can we have async agentic flows locally?
All the best, fellas!