Run your AI agents
in parallel.
LaboLabo runs Claude Code, Codex, and Gemini side by side — one git worktree per session — with live git state and diffs right next to a real terminal.
brew install --cask love-rox/tap/labolaboOne window, the whole loop
Kick off work, watch it happen, review the diff, and ship — without juggling terminals and tabs.
Run a session per task
A bugfix here, a feature there, a spike in a third repo. Each session is its own git worktree, so agents editing at the same time never collide. The sidebar lists them all, grouped and color-coded by repository, each with a live status dot — running, waiting for input, or idle.
See every change as it lands
A real commit graph tracks branches, tags and ahead/behind while the agent works. Changed files and a red/green diff update the instant an edit happens — toggle any file between diff and full contents.
A real terminal, embedded
A genuine libghostty GPU terminal runs the agent interactively — the true thing, not a pale imitation.
Status without scraping
Running, waiting for input, idle — session state comes from Claude hooks over a socket. Accurate, not guessed.
Resume across restarts
Quit and reopen — sessions come back with --resume, so a crash or a reboot never loses your place.
Ship it from the app
When the agent is done, review the diff, then push and open a PR with gh — without leaving the window.
Any AI CLI
Claude Code, Codex, Gemini — swappable adapters, each surfacing exactly the capabilities it supports.
Native macOS
SwiftUI and Swift 6. No Electron. Fast, light, and completely at home on your Mac.
Install in one line
Ships via a Homebrew cask. Free and ad-hoc signed, so the first launch needs a quick approval in Settings › Privacy & Security.
Or download the .app