Git worktree vs branch: what's the difference and when to use each

What a branch is

A branch is a pointer, a file in .git/refs/heads/ holding one commit ID, that moves forward when you commit. Creating one costs nothing and changes no files. In a normal clone you see one branch at a time: git switch rewrites the files in your folder to match the branch you switch to.

What a worktree is

A worktree is an extra working directory attached to the same repository. It has its own files, its own index and its own HEAD, but shares commits, branches and config with the main checkout. With worktrees you can have several branches checked out at once, each in its own folder.

Side by side

	Switch branches	One worktree per branch
Folders on disk	One	One per branch
Branches open at once	One	Many
Uncommitted work when switching	Must commit or stash first	Stays put in its folder
Dependencies and build cache	Rebuilt after many switches	Installed once per folder
Disk use	Lowest	Extra checked-out files, plus each folder’s dependencies
Same branch in two places	n/a	Not allowed
Good for	Short, sequential tasks	Parallel work, long builds, AI agents

When switching branches is enough

• You work on one thing at a time and finish it before starting another.
• The project builds fast and has no heavy per-branch dependencies.
• You rarely have uncommitted work when you need to switch.

When a worktree is better

• Interruptions. A hotfix comes in while you’re mid-feature. Open a worktree on main instead of stashing half-done work.
• Reviewing a pull request while your own branch keeps its state, running dev server included.
• Long-running jobs. A test suite or build runs in one folder while you keep coding in another.
• AI coding agents. Claude Code, Codex and others edit files directly, so each needs its own folder. See running agents in parallel and claude --worktree.

What about stash and clone?

Stash parks uncommitted changes so you can switch branches in the same folder. It works for a quick detour, but stashes are easy to forget, and they are shared by every worktree of a repository. A second clone gives you another folder too, but with a separate copy of the history. You have to fetch and push between clones, and branches in one clone aren’t visible in the other. A worktree gives you the separate folder without the separate repository.

The catch with worktrees

Folders are easy to make and easy to forget. A month in, you might have a dozen worktrees: some merged, some with commits that only exist on your laptop, some with changes nobody committed. Orchard shows every worktree on your Mac in one view and says which ones still hold work and which ones are already in main. Cleaning up by hand? Use our cleanup checklist.

FAQ