Every model you have
Each teammate runs on the AI and model you pick, from the lists your own coding agents offer. Set the effort, switch a conversation to another AI partway through, and run up to eight at once.
Put Claude Code, Codex, Cursor, Copilot, OpenCode, Antigravity and Muse Code side by side, give each teammate its own model, and watch them work on your machine, on the accounts you already have. Nothing is pooled, proxied or resold. No plan yet? Start free with OpenCode.
Each teammate runs on the AI and model you pick, from the lists your own coding agents offer. Set the effort, switch a conversation to another AI partway through, and run up to eight at once.
Give one task to two or three models side by side, each in its own copy of the folder, or blind with the names hidden until you pick. Keep the one you like. See three models build the same game →
In Approve each, nothing runs until you have seen the exact command or change and where it will happen. Claude Code asks before it uses a connected account.
Every run is written to a local record as it happens. Close the app mid-run and the run is kept, ready to pick back up.
Teammates hand work to each other by name, and a hand-off chain puts a checker on a different AI that must approve before the run counts. The Board shows every conversation by what it needs from you.
Run a teammate every few hours, daily, weekly, once, or when files in a folder change, and keep it running in the background after the window is closed.






Windows 10 or 11, 64-bit
x64 · 118 MB
Unsigned — SmartScreen will warn on first install
Locust drives one; it does not bundle one
First launch offers to install what is missing
Locust carries its own npm, so no Node.js needed
macOS 12 or later · Apple silicon or Intel
Not notarized — the first open asks you to allow it in System Settings › Privacy & Security
Updates itself, like on Windows · a new Mac build every few releases
Locust-Setup.exe
Published with each release on GitHub