How do I start Claude Code in a git worktree?

Run claude --worktree (or -w). Claude Code creates a worktree under .claude/worktrees/ / on a new branch named worktree- and starts the session in it.

Are my .env files in a new worktree?

No. A worktree is a fresh checkout, so gitignored files are missing. Add a .worktreeinclude file listing them, in .gitignore syntax, and Claude Code copies matching gitignored files into each new worktree it creates.

What happens to the worktree when I exit?

If it is clean, an unnamed session removes it and its branch automatically, and a named session asks first. If it has changes or new commits, Claude Code asks whether to keep or remove it.

Guide

Claude Code worktrees: run parallel sessions without collisions

By Eren Çolak, who makes Paneherd · Updated 24 October 2026

Run claude --worktree feature-auth and Claude Code creates a separate checkout for that session under .claude/worktrees/feature-auth/, on a new branch named worktree-feature-auth, then starts in it. Run the same command with another name in a second terminal and the two sessions cannot overwrite each other's files. This guide follows Anthropic's worktrees documentation (checked 24 October 2026); versions differ, so check yours.

Start one

claude --worktree feature-auth
claude -w bugfix-123      # short form, in another terminal

The repository needs at least one commit. If you leave out the name, Claude generates one such as bright-running-fox. By default a new worktree branches from the repository's default branch on the remote, so it starts clean; set worktree.baseRef to "head" in settings to branch from your current work instead. To start from a pull request, quote the reference: claude --worktree "#1234".

Add .claude/worktrees/ to your .gitignore so the folders do not show up as untracked files in your main checkout.

Set up the environment

A worktree is a fresh checkout: dependencies are not installed and gitignored files such as .env are missing. Ask Claude to install dependencies, and list files to carry over in a .worktreeinclude file at the project root:

.env
.env.local
config/secrets.json

It uses .gitignore syntax, and only files that match and are also gitignored are copied. Each worktree also needs its own dev server port if you run servers.

What happens when you exit

Clean worktree: an unnamed session removes it and its branch. A named session asks first.
Changes or new commits: Claude asks whether to keep or remove it. Keeping preserves the folder and branch; resume later with the claude --worktree <name> --resume command it prints. Removing deletes the work.

To remove one yourself: git worktree remove <path> (add --force if it has changes; run git worktree unlock first if git says it is locked). Subagents can have their own worktrees too: add isolation: worktree to a custom subagent's frontmatter.

What worktrees do not isolate

They separate file edits. All worktrees still share one repository .git directory and your plan's usage limits: each extra session spends the same allowance (details). You also still have to watch each session separately.

Seeing all of them at once

Paneherd opens one terminal per session in a grid in VS Code or Cursor, so you can run a worktree session in each pane and see which is working, waiting or finished from one sidebar. It works with --worktree rather than replacing it. See how to run multiple Claude Code sessions at once.

Frequently asked questions

Do I need worktrees for two sessions?

Only if both change code in the same repository. Read-only sessions or sessions in separate projects do not collide.

Can two sessions use the same worktree name?

Passing a name whose folder already exists opens that existing worktree rather than creating a new one.

Install Paneherd free