stablyai/orca

Orca is a desktop application for people who run several AI coding agents at once, such as Claude Code, Codex, or OpenCode, and want a single place to manage all of them instead of juggling separate terminal windows. It calls itself an orchestrator, meaning its job is to organize and coordinate multiple agents rather than being a coding agent itself. The core idea is running each agent in its own git worktree, which is a separate working copy of your code that shares the same repository. This lets you send the same task to several agents in parallel, each working in isolation, and then compare their results side by side before choosing which one to merge. Orca also includes full terminal support with fast rendering and scrollback that survives restarts, so you can still work directly in a shell when needed. Beyond running agents, Orca adds tools meant to speed up reviewing their work. A design mode lets you click any element in an embedded browser window and send its HTML, styling, and a screenshot straight into an agent's prompt. You can annotate lines in a code diff with comments that get sent back to the agent, drag files or images directly into a prompt, and browse GitHub and Linear issues and pull requests without leaving the app. There is also a mobile companion app so you can check on running agents and send follow-up instructions from your phone, plus an SSH mode for running agents on a remote server instead of your local machine. Orca is built with TypeScript and is available for macOS, Windows, and Linux, with users bringing their own subscriptions to whichever coding agents they use. It also ships a command line interface so agents themselves can script Orca, and the project is released under the MIT license.