using-git-worktrees

You can isolate risky refactors by creating a git worktree on a new branch, which keeps your main workspace clean and recoverable if the plan leaves the code in a broken state. Work happens in a separate directory under .claude/plugin/worktrees/.

Use git worktrees for parallel feature development when running multi-step automated workflows, ralph loops, or risky refactors that might leave the branch broken and require a rollback point. They prevent workspace contamination across concurrent tasks.

Clean up a git worktree after merging by running git worktree remove on the isolated directory, then deleting the corresponding branch. This safely removes both the working directory and the branch created for the isolated task.

Managing branches during isolated feature work is best handled by creating a dedicated branch like dojo/<task-slug> with a corresponding worktree, committing changes inside that isolated directory, then chaining to merge or PR workflows for integration.

Yes, you can run automated workflows in a git worktree without affecting your current branch. The worktree creates an isolated working directory on a separate branch, so any broken state from the workflow stays contained and recoverable.

Git worktrees provide full workspace isolation for refactors, whereas stashing only temporarily shelves changes. Worktrees allow you to keep your main workspace active while working on a separate branch, making them safer for multi-step plans that might break code.