Run Claude Code, Codex, Cursor CLI, OpenCode or any shell side by side, each in its own Git worktree, with a queue of agents waiting on you and diff review. A separate daemon holds the sessions, so agents keep running when you quit, crash or update the app. Rust desktop app, no Electron or web view. Oct 28 benchmark (Krill 0.2.20 vs Orca 1.4.242, 10 terminals, screen on): 1/7 the memory; slower on big output bursts. Free, Apache-2.0. Apple Silicon Mac; Windows preview.

Hi PH, I'm Owen. I usually have ~10 Claude Code / Codex sessions running at once, each in its own git worktree, and the Electron app I used for that was the heaviest thing on my Mac. So I rebuilt the same workflow as a native app (Rust + gpui, no Electron, no web view).

In our 10-terminal benchmark (Oct 25, Krill 0.2.2 vs Orca 1.4.237) Krill used 86 MiB vs 1,275 MiB. It isn't faster everywhere: in a 20-terminal output test it was slower (7.5 s vs 4.0 s). Method and raw numbers are in the README.

Honest limits: the macOS build isn't notarized. The install script verifies the checksum and signature and then removes quarantine, so please read it first or build from source if you prefer. Windows is a preview.

If you run 5+ agents at once, I'd love to hear what breaks first.

Krill

Run parallel AI coding agents natively in Rust, not Electron

Run parallel AI coding agents natively in Rust, not Electron