If you run multiple AI agents on the same repo, how do you stop them stepping on each other?
Heat trend
Collecting trend data
The percentage is based on available heat signal, not comment count or independent people.
A developer uses multiple AI agents in parallel to develop 2 or 3 features simultaneously on the same repository, separating code with git worktrees.…
I use AI to develop, and because it works quite well, I extended it to run multiple agents in parallel. This way I can develop 2 or 3 features at the same time. It works well, and git worktrees do the job of separating the code.
The problem comes when I try to test, or when I ask them to test against a running app. They can't all use the same db and running instance. I still have to do that part manually, one by one, which is annoying and slows the whole thing down.
Curious whether others actually run agents in parallel on one repo, or just do them one at a time to avoid the mess. And if you do run them in parallel, how are you keeping both the code and the data from colliding, and how are you testing them properly? Or is this just me overcomplicating it?