Intro
I am using AI such as Claude Code (You're absolutely right!), Codex, Cursor, and Cline for both work and for fun.
・My app DEV post (🚀How I released Chrome Extensions (💪🧠Muscle Brain v4))
・My website DEV post (🧐I created a website animation that you might stare at for a while (GSAP)🎨)
I am always thinking about how to use AI better.
・🤖How to make AI follow your instructions more for free (OpenSpec)📝
・🧠How to use AI more efficiently for free (Serena MCP)🧐
Running AI in parallel seems to be one good way to use AI efficiently. When running AI in parallel, I first came up with Git Worktree. However, it looked complicated and difficult, so I was waiting and seeing for a while. But recently, I heard that Git Worktree Runner can easily run AI in parallel. So, I finally got off my butt and gave it a try.🚀
One AI Coding. (Without Git Worktree Runner🙅) ↓
Parallel AI Coding. (With Git Worktree Runner🙆) ↓
Super Parallel AI Coding. (Maybe in the future?😆) ↓
What is Git Worktree Runner?
Usually, we can use only one Git branch at a time, and we have to switch to another branch to use others. However, Git Worktree allows us to use many Git branches at once. For this reason, Git Worktree is often used when running AI in parallel.
To explain in short, Git Worktree Runner is a wrapper for Git Worktree for easy use. You can use Git Worktree with short and simple commands with less stress. Git Worktree Runner is made by the famous AI PR request review service company CodeRabbit🐰, so it is quite reliable. Moreover, it is an OSS, so you can use it for free.🤑
You can use Git Worktree Runner with the tools below. ↓
| Environment | Support |
|---|---|
| Editor | Cursor, VS Code, Zed, and more |
| AI tool | Claude Code, Codex, or other |
| Platform | Windows (Git Bash), Linux, macOS |
GitHub URL of Git Worktree Runner ↓
https://github.com/coderabbitai/git-worktree-runner
How to use Git Worktree Runner
There are only 4 main commands to use Git Worktree Runner: new, edit, ai, and rm.
| Command | Description |
|---|---|
new |
Create worktree |
editor |
Open editor |
ai |
Use AI |
rm |
Remove worktree |
Let's take a look at how to use Git Worktree Runner step by step.😄
Requirements
・ Git 2.5+ (for git worktree support)
・ Bash 3.2+ (macOS ships 3.2; 4.0+ recommended for advanced features)
Installation: ↓
I will explain for Windows Git Bash, so please check the official documentation for other installation methods.
git clone https://github.com/coderabbitai/git-worktree-runner.git
cd git-worktree-runner
complete -p git
# Check whether "complete" is installed.
# Usually "complete" is installed with Git, but if not install it.
echo 'export PATH="$PATH:'$(pwd)'/bin"' >> ~/.bashrc
source ~/.bashrc
# This creates a symbolic link and makes "git-gtr" executable.
# This is often said as "Add to PATH".
Setup: ↓
cd ~/your-repo # Navigate to Git repo
git gtr config set gtr.editor.default vscode # One-time setup
git gtr config set gtr.ai.default claude # One-time setup
Use: ↓
git gtr new my-feature # Create worktree
git gtr editor my-feature # Open editor
git gtr ai my-feature # Start AI
git gtr rm my-feature # Remove when done
Hooks: (Optional) ↓
You can use Hooks to run custom commands after worktree operations. Here are examples.
# Post-create hooks (multi-valued, run in order)
git gtr config add gtr.hook.postCreate "npm install"
git gtr config add gtr.hook.postCreate "npm run build"
# Post-remove hooks
git gtr config add gtr.hook.postRemove "echo 'Cleaned up!'"
There are many other commands and their options, so please run git gtr help or check the official documentation for full information. ↓
https://github.com/coderabbitai/git-worktree-runner
There are several patterns for using AI in parallel:
・Frontend, Backend, Test, etc.,
・A part, B part, C part, etc.,
・Code same thing with different AI,
And any other pattern you can think of, so let's find our own best way.
Outro
I’m using both Claude Code and Codex, but Codex seems to keep me waiting for a while. So running AI in parallel with Git Worktree Runner might be a good way to use time more effectively.
I hope you learned something from this post.
Thank you for reading.
Happy AI coding!🧠
You're absolutely right!🤖
If you have any good ideas or know anything about running AI in parallel, please leave a comment below.⬇️⬇️
Top comments (1)
Nice write-up 👍 — this “git worktree + AI” combo is one of those quiet shifts that really changes how you work. Love how you break down the parallel-workflow benefits step by step.
Working around non-tech people (fintech, finance, biz folks) reminds me how powerful this can be beyond just code: it means fewer pauses, fewer “wait—lemme stash” moments, and more fluid work.
Thanks for sharing — tools like this make you wonder why nobody told us earlier 🤯