Level 1
A linked worktree
Another checkout of the same repository. Isolates files. Carries nothing else.
Parallel agents in lpm
Git worktrees for Claude Code, Codex, and any coding agent.
Two agents in one folder overwrite each other. A worktree fixes that in one command — and then leaves you to rebuild the environment around it. This page covers what worktrees do for agents, what Claude Code and Codex now do natively, and the three levels of isolation you can choose between.
Jump to the comparison
Level 2
A managed worktree
The same checkout, created and torn down for you, with the agent already running in it.
Level 3
A standalone copy
The project as it exists on disk right now — local files, dependencies, uncommitted work — with its own Git repository.
The short answer
How do you run multiple AI coding agents on one repository?
Give every agent its own directory. A Git worktree is the cheapest way to do it: a second working directory on its own branch, backed by the same repository history.
One worktree per agent
git worktree add ../app-auth -b auth
cd ../app-auth
claudeClaude Code and Codex now do this for you: claude --worktree creates, enters, and cleans up a worktree per session, and the ChatGPT desktop app gives each Codex chat one. The rest is still yours to set up: copying your .env, installing dependencies, carrying over uncommitted work, and sharing the ports and databases every agent still uses.
That gap is why a second primitive exists. lpm creates worktrees too, and lpm Duplicate can instead copy the project as it sits on your disk into a standalone project with its own Git repository — then start the agent in each one.
Before you fan out
Five things a Git worktree does not carry
Creating the worktree is one command. These are the five reasons the agent inside it still cannot run your project.
Your .env and other ignored files are gone:
What you see Every environment variable is undefined
Fix in a worktree List them in .worktreeinclude, or copy them by hand
In a copy Comes along
- Why
- The agent starts, the app boots, and every environment variable is undefined.git worktree add checks out tracked files only. Anything in .gitignore — .env, .env.local, local certificates, editor settings — was never in the commit, so it is never in the worktree.
- What fixes it
- Claude Code and Codex both read a .worktreeinclude file and copy the gitignored files that match it into the worktrees they create. For everything else you copy them yourself, or start from a copy of the project instead of a checkout.
node_modules is empty in every new worktree:
What you see A full install per agent before any work starts
Fix in a worktree Install per worktree, with a setup script
In a copy Comes along
- Why
- A full install per agent, and another copy of every package on disk, before any work starts.Dependencies are ignored files too. A fresh checkout has none of them, and symlinking one shared directory breaks resolution in tools that walk the real path.
- What fixes it
- Install per worktree and accept the cost, with a setup script if your agent runs one, or duplicate the project so the already-installed dependencies come with it.
The work in progress on your desk does not come along:
What you see The agent starts from a commit you moved past
Fix in a worktree Commit or stash first
In a copy Comes along
- Why
- You ask an agent to continue what you were doing, and it starts from a commit you moved past hours ago.A worktree branches from a commit. Claude Code branches from your default branch by default, or from local HEAD when worktree.baseRef is set to head. Neither reproduces uncommitted edits or untracked files.
- What fixes it
- Commit or stash first, or copy the working project rather than checking one out.
Two agents cannot try the same branch:
What you see
fatal: 'main' is already checked outFix in a worktree One branch per agent
In a copy Comes along
- Why
fatal: 'main' is already checked out at '/path/to/repo'Linked worktrees share one repository, and Git refuses to check out a branch in two of them — a commit in one would leave the other pointing at a stale state of the same branch.- What fixes it
- Give each agent its own branch, or give each one an independent repository so the restriction no longer applies.
Ports, databases, and Docker volumes are still shared:
What you see The second dev server can't bind its port
Fix in a worktree Per-copy config, separate services or containers
In a copy Still shared
- Why
- The second dev server cannot bind. Two agents run migrations against one database and corrupt each other's fixtures.Every isolation model on this page draws its boundary at the filesystem. Nothing about a separate directory reserves a port, namespaces a Postgres schema, or forks a Docker volume.
- Where this stands
- No worktree model solves this, and lpm does not either. What lpm does is catch it: it checks declared ports before a project starts, tells you which process is holding one, and can free it or stop the start. Real runtime isolation still means per-copy configuration, separate services, or containers.
The first four go away when you copy the whole project instead of a worktree: your .env, installed dependencies, and uncommitted work come along by default, and the copy has its own repository, so it can stay on a branch that is checked out elsewhere. The fifth holds for copies too.
What the agents already do
Claude Code and Codex create worktrees for you
Before adding a tool, check what your agent ships with. For a single session on a clean branch, the built-in flag is usually enough. Checked against each CLI's documentation in September 2026.
Claude Code
Built inPass --worktree (or -w) with a name and Claude creates the worktree, starts the session inside it, and offers to remove it on exit. Worktrees land under .claude/worktrees/<name>/ on a branch named worktree-<name>. Subagents can get the same treatment with isolation: worktree in their frontmatter.
claude --worktree feature-auth
# copy gitignored files into every new worktree
cat .worktreeinclude
.env
.env.localNew worktrees branch from your default branch unless you set worktree.baseRef to head. Anthropic's own documentation is explicit that a worktree is a fresh checkout and you still have to initialize the environment inside it. Claude Code worktree documentation
Codex
Built inCodex in the ChatGPT desktop app can give each chat its own worktree, so several run in the same project in parallel. Worktrees are created under $CODEX_HOME/worktrees/ in a detached HEAD state rather than on a branch, leaving your main checkout untouched, and older ones are cleaned up automatically.
It reads a .worktreeinclude file too, and a local environment can run a setup script in each new worktree. Uncommitted work, ports, and databases are still yours to handle. Codex worktree documentation
GitHub Copilot, Gemini CLI, OpenCode, and the rest
Bring your ownWorktrees are a Git feature, not an agent feature, so any terminal agent works inside one. What varies is whether the agent creates and cleans up the directory for you, or whether you script it.
git worktree add ../app-copilot -b copilot-task
cd ../app-copilot
cp ../app/.env .env
pnpm install
copilotThis is the loop that agent-specific flags remove — and the loop that stays if the agent you use does not have one.
Isolation models
Five ways to isolate a coding agent, compared
Raw Git, the worktrees Claude Code and Codex create for you, and both lpm primitives, side by side, so you can see exactly where each boundary is drawn.
Isolated working directory
- All five
- Yes
Git repository
- git worktree
- shared
- claude --worktree
- shared
- Codex worktrees
- shared
- lpm Worktree
- shared
- lpm Duplicate
- independent
Created and removed for you
- git worktree
- No
- claude --worktree
- Yes
- Codex worktrees
- Yes
- lpm Worktree
- Yes
- lpm Duplicate
- Yes
Carries .env and other ignored files
- git worktree
- No
- claude --worktree
- .worktreeinclude
- Codex worktrees
- .worktreeinclude
- lpm Worktree
- No
- lpm Duplicate
- Yes
Carries installed dependencies
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- setup script
- lpm Worktree
- install
- lpm Duplicate
- Yes
Starts from your uncommitted work
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- No
- lpm Worktree
- No
- lpm Duplicate
- Yes
Two agents on the same branch
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- No
- lpm Worktree
- No
- lpm Duplicate
- Yes
Create many at once
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- one per chat
- lpm Worktree
- 1–50
- lpm Duplicate
- 1–50
Queue the same prompt on each
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- No
- lpm Worktree
- Yes
- lpm Duplicate
- Yes
Inherits the project's services and actions
- git worktree
- No
- claude --worktree
- No
- Codex worktrees
- No
- lpm Worktree
- Yes
- lpm Duplicate
- Yes
Works with any terminal agent
- git worktree
- No
- claude --worktree
- Claude only
- Codex worktrees
- Codex only
- lpm Worktree
- Yes
- lpm Duplicate
- Yes
Isolates ports, databases, Docker volumes
- All five
- No
The last one is worth reading twice. Nothing above reserves a port, namespaces a database, or forks a Docker volume, including both lpm entries. Filesystem isolation is where all five stop, and shared ports and databases are where parallel agents usually collide. lpm catches the port half of it when a project starts: it tells you which process holds a declared port and can stop it or cancel the start. Giving each copy its own ports is up to your configuration.
One prompt, three agents
Both primitives, the same fan-out
lpm does not ask you to give up worktrees. Pick the kind from the project menu, and both share the same fan-out form for queuing the work.
lpm Worktree
Linked worktrees, created in a batch
Each one is a real Git worktree on a new branch from your current commit, sharing the repository. Uncommitted and ignored files stay behind, the same as raw Git. Tick Install dependencies when the worktree needs them.
- Deleting one also deletes its branch, so merge or push anything you want to keep first
- Needs a Git repository with at least one commit, with the project at its root
lpm Duplicate
Standalone copies of the project you have now
A copy of the folder with its own Git repository, carrying uncommitted work, ignored files, and installed dependencies; on APFS it is a copy-on-write clone. Regenerable build caches are left behind.
- Pulls the latest commits by default; optionally keep committed work only or reinstall dependencies
- Each copy is independent, so several can sit on the same branch
- 01
Right-click a project and choose Duplicate for standalone copies or New worktree for linked worktrees, then set how many, up to 50.
- 02
Label them, group them in the sidebar, and pick the action or command each one should run, with the prompt to send. Each copy can get its own.
- 03
Watch every copy's agent status from the sidebar, review the diffs, and delete the copies you do not keep. Deleting is permanent.
The agents can drive this themselves. lpm installs skills for Claude Code, Codex, Gemini CLI, and OpenCode, so an agent asked to try three approaches can create its own copies, run the work in them, wait for the others to settle, and clean them up when you have merged the one you want. Already typing the prompt? Choose Run in duplicates from the composer’s send menu to run it here and in fresh copies at once, or read the full guide to running Claude Code in parallel.
Pick one
Which isolation model do you actually need?
Work down the list and stop at the first line that describes your repository.
One agent, one clean branch, and setup is cheap
Your agent's built-in flag
claude --worktree, or a worktree chat in the ChatGPT desktop app for Codex. Nothing to install, and the agent cleans up after itself.
Several agents, and your repository needs no local setup
Git worktree, or lpm Worktree for the batch
The checkout is all the agent needs, so the lightest primitive wins. Use lpm Worktree when you want several at once with a prompt queued on each.
The agent needs your .env, your dependencies, or your uncommitted work
lpm Duplicate
A checkout cannot reproduce state that was never committed. A standalone copy of the working project can.
Several agents should attempt the same branch and you keep the best
lpm Duplicate
Linked worktrees refuse a branch that is already checked out. Independent repositories have no such restriction.
You are running agents with approvals turned off
A container or a VM
None of these models is a security boundary. A separate directory does not contain a command that decides to touch the rest of your machine.
FAQ
Git worktrees and AI agents, answered
Does Claude Code support Git worktrees?
Yes. Pass --worktree or -w with a name and Claude Code creates the worktree, starts the session in it, and offers to remove it when you exit. Worktrees are created under .claude/worktrees/<name>/ on a branch named worktree-<name>, and subagents can be pinned to their own worktree with isolation: worktree in their frontmatter.
Does a Git worktree copy files like .env or node_modules?
No. git worktree add checks out tracked files from the commit you point it at. Anything ignored by Git — .env files, node_modules, virtualenvs, local certificates — is absent from a new worktree. Claude Code and Codex can copy selected ignored files into the worktrees they create if you add a .worktreeinclude file, and lpm Duplicate carries them because it copies the project folder instead of checking one out.
Should every Claude Code session get its own worktree?
Only if two sessions could touch the same files. A worktree hands each session a separate checkout on its own branch, so their edits cannot collide; sessions working in unrelated parts of the repo can share one checkout. Whatever directory you pick — a worktree, a standalone copy or a container — sessions on one Claude account still share its plan limits, so the ceiling is usually how many diffs you can review.
Can two Git worktrees check out the same branch?
No. Git refuses with 'fatal: <branch> is already checked out at <path>', because a commit in one worktree would leave the other pointing at a stale state of the same branch. If you want several agents to attempt the same branch and then keep the best result, each one needs an independent repository rather than a linked worktree.
Does Codex support parallel agents and worktrees?
Yes. Codex in the ChatGPT desktop app can run each chat in its own Git worktree, created under $CODEX_HOME/worktrees/ in a detached HEAD state, so several chats work in parallel and your main checkout stays untouched. It can copy files listed in .worktreeinclude and run a setup script in each worktree, but as with Claude Code, ports and databases are still shared.
Do Git worktrees isolate ports, databases, or Docker volumes?
No, and lpm does not isolate them either. Every model on this page draws its boundary at the filesystem, so two agents in two worktrees will still fight over port 3000 and still run migrations against the same database. lpm checks declared ports before a project starts and tells you which process is holding one, so the collision surfaces immediately instead of halfway through a run. Isolating runtime state needs per-copy configuration, separate services, or containers.
Can I use Git worktrees with GitHub Copilot, Gemini CLI, or OpenCode?
Yes. Worktrees are a Git feature, so any terminal agent runs inside one. What differs is whether the agent creates and cleans up the directory for you. Claude Code and Codex do; for the others you run git worktree add yourself, or use a tool that does the batch for you.
Is there a Git worktree MCP server or agent skill?
lpm installs skills for Claude Code, Codex, Gemini CLI, and OpenCode that teach the agent how to drive lpm directly. An agent can then create its own isolated worktrees or copies, run work in them, wait for the others to settle, and remove them, without you translating each step yourself.
What is the difference between lpm Worktree and lpm Duplicate?
They are separate items in a project's right-click menu. New worktree creates real linked Git worktrees, each on a new branch from your current commit, sharing your repository. Duplicate creates standalone folders with their own Git repository, starting from the project exactly as it is on disk. Both create up to 50 at a time, inherit the project's services, actions, and pinned Claude account, and can queue an action or command with a prompt on each one.
Does lpm Worktree copy my .env file and dependencies?
No. lpm Worktree creates a real Git worktree, so it has the same blind spot as raw Git: uncommitted and ignored files are not carried over. Tick Install dependencies to install them fresh in each worktree. If the copy needs your local files and current state, use lpm Duplicate instead.
What happens when I delete an lpm worktree or copy?
lpm deletes the folder outright and skips the Trash, and removing a worktree also force-deletes its branch, so push or merge anything worth keeping first.
How much disk does each copy use?
A linked worktree is very compact because the repository data is shared. On APFS, a Duplicate begins as a clone whose unchanged blocks stay shared on disk, and it only grows as each copy changes; where the disk can't clone, as on ext4 or Windows, it is a full copy. Regenerable build caches such as .next, dist, and target are skipped rather than cloned.
How many parallel agents should I run?
Fewer than you can create. Spinning up five agents takes one command; reviewing five diffs and landing five branches does not scale the same way, and every session draws on the same plan limits. The bottleneck is usually review, not isolation.
Keep reading
Git worktree alternative
The standalone-copy side of the comparison in full: what Duplicate carries that a checkout cannot.
Best terminal for Claude Code & Codex
Run AI coding agents side by side with each project’s services, logs, and status in view.
Connect AI agents to your dev environment
Give agents a CLI to run services, read logs, wait for readiness, and fan out into project copies.
Run Claude Code in parallel
The whole workflow for several Claude Code sessions at once, from one prompt to the diff you keep.
Run Claude Code on a remote server
Fan out on a Linux box with cores to spare instead of the laptop you are typing on.
A cmux alternative for parallel agents
cmux gives every agent its own terminal tab; see how that compares with giving every agent its own checkout.
Worktrees when a checkout is enough.
Copies when it is not.
Download lpm and fan one prompt out to Claude Code, Codex, or any terminal agent — in linked worktrees or standalone project copies, whichever the task needs.