Some git worktree thoughts

Okay, so everyone's talking about worktrees, and I wanted to share my perspective. I'm not using them yet, and that's alright. I don't want people to feel pressured to use them just because they see others doing it on Twitter. I'm getting plenty done without them.

Basically, a worktree is a built-in Git feature that lets you have multiple copies of your project that all share the same underlying Git objects. People are using them to work on multiple things in a single repository simultaneously. If that works for them, great, but I'm not there yet.

What I've been doing instead is creating separate folders and Git cloning my project into each of them. For example, I have desktop, desktop02, and desktop03. These aren't worktrees; they're just separate checkouts. This lets me work on different features simultaneously. For example, I can work on the redesign in desktop03 while addressing bug fixes in desktop-main.

While worktrees seem to have some potential gotchas that make me nervous about corrupting my Git state, I've found this approach helpful. If you have different tasks in the same repo, this might be a good step before diving into full worktrees. If you're using worktrees, that's fantastic.

My advice is to stay involved in the process and keep an eye on what's happening. Tools and assistants can still make mistakes. If you find that your agents are taking a long time to investigate something, instead of getting distracted, try using discrete checkouts. This lets multiple agents work without interfering with each other, without fully committing to worktrees, which I find quite sophisticated, and I'm just not there yet.