Skip to content

Guide · Open source

How to use Orca: run several coding agents side by side

Updated · written with AI from the project's own documentation and checked against our catalogue data

Orca is an open-source desktop application, with a mobile companion, for working with a group of coding agents at the same time. It is aimed at developers who already use command-line coding agents and want to run several of them on one project. After this guide you will know what Orca does, how to get the app, and how to run and compare agents in parallel worktrees.

Orcastablyai/orca · ★ 113k · MIT

Quick start

  1. 1Download the app. Get the desktop build for your system from the Orca releases page.
  2. 2Open Orca. Launch the desktop app on macOS, Windows or Linux.
  3. 3Connect your agents. Run the CLI coding agents you already use, such as Claude Code, Codex, OpenCode or Pi, with your own subscription.
  4. 4Create parallel worktrees. Send one prompt across several agents so each works in its own isolated git worktree.
  5. 5Compare and merge. Review the results side by side, then keep and merge the one you want.
  6. 6Annotate diffs. Comment on diff lines and send them back to the agent before committing.

What Orca is and who it is for

Orca is described as the place for working with a fleet of agents, such as Grok Bot or Claude Code. The project's README says you can run any coding agent with your own subscription. The README's example list includes Codex, ClaudeCode, OpenCode and Pi, and it says any CLI agent that runs in a terminal runs in Orca.

Orca fits developers who already use CLI coding agents and want to run several at once, people who want to compare outputs from different agents on the same prompt, and developers who work on remote machines. It is not a good fit if you do not use coding agents, because the core value depends on running them.

  • Licence: MIT
  • Language: TypeScript
  • Platforms: macOS, Windows and Linux builds are listed for the desktop app, with a separate mobile companion

Install the Orca desktop app

The README's install method is to download the desktop app from the releases page, which lists macOS, Windows and Linux builds. The project's download page is linked from the README header. The README lists no separate prerequisites.

The README also includes a git clone command for the repository. Use it if you want the source code, which is written in TypeScript.

  • Open the releases page linked from the README and choose the build for your system.
  • Install and open the app.
  • Optionally, get the source with the clone command below.
git clone https://github.com/stablyai/orca

Run parallel agents in isolated worktrees

Orca's parallel worktrees feature lets you fan one prompt out across several agents. Each agent gets its own isolated git worktree, and everything is tracked in one place. You can then compare the results and keep the one you want.

Because each agent works in a separate worktree, their changes do not overwrite each other while they run. You review the outputs side by side and merge the one you choose.

  • Start agents such as Claude Code, Codex, OpenCode or Pi in the same session.
  • Give them the same prompt to compare their approaches.
  • Keep the winning result and merge it.

Review and annotate AI-generated diffs

Orca's Annotate AI Diffs feature lets you drop comments on any diff line and send them back to the agent. You can review, edit and commit without leaving Orca.

This suits teams that review AI-generated diffs before committing them.

Use Orca with remote machines over SSH

The SSH Worktrees feature runs agents on a remote machine with file editing, git and terminals. The README says auto-reconnect and port forwarding are included.

Use this if you work on remote machines and want agents running there.

Monitor agents from your phone

The mobile companion lets you monitor and steer your agents from your phone. You can get notified when an agent finishes and send follow-ups from anywhere. The README links an iOS App Store listing and an Android APK download, with docs at onorca.dev.

What people use it for

  • Running Claude Code and Codex on the same prompt and comparing their diffs.
  • Keeping several agent tasks running in separate worktrees on one repository.
  • Running agents on a remote box over SSH while editing files locally.
  • Checking agent progress and sending follow-ups from a phone.

Frequently asked questions

Is Orca free?

The README states that Orca is released under the MIT licence. It says you run agents with your own subscription, so any agent subscription costs are separate from Orca.

What do I need to install Orca?

The README lists no prerequisites for the desktop app. You download a build for macOS, Windows or Linux from the releases page.

Which coding agents does Orca support?

The README says Orca works with any CLI agent that runs in a terminal. Its examples include Claude Code, Codex, OpenCode, Pi and Grok.

Can I run more than one agent at a time?

Yes. The parallel worktrees feature fans one prompt across several agents, each in its own isolated git worktree, so you can compare their results.

Is Orca a browser-based tool?

No. The README describes a desktop application with a separate mobile companion, so it is not a fully browser-based tool.

Are there alternatives to Orca?

The source does not name comparable projects, so this guide cannot make a comparison. Check the README and releases page for current details.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Orca suits developers who already use CLI coding agents and want to run several at once, compare their outputs, or work over SSH on remote machines. It is a poor choice if you do not use coding agents or need a browser-only tool. The README warns that its changelog is the real feature list and changes often, so check it before relying on any specific feature.

Install options & downloads