What is Orca CLI? Install and Run the First Commands

Orca CLI is the orca command that scripts a running Orca desktop app (an agent development environment, ADE). It is not a model. It orchestrates agents you already use — Claude Code, Codex, Cursor CLI, and other terminal agents — each in its own worktree. Install from the official Install page. First checks are command -v orca and orca status --json.

This note stays on documented install, registration, and verify steps. No prices or personal reviews. If flags change, prefer CLI overview and CLI reference.

How is Orca different from a regular IDE?

One-line answer: A regular IDE is an editor for one person. Orca is an ADE that gives each task its own git worktree, agent terminal, and browser tab.

What is Orca? describes a desktop tool for running multiple coding agents side by side. Each task gets its own git worktree, agent terminal, and browser tab, so you can fan out Claude Code, Codex, and Cursor CLI without stashing or branch-juggling.

The docs also state what Orca is not:

  • Not a model. Bring your own Claude, Codex, or OpenCode subscription.
  • Not a git replacement. Every worktree is a real git worktree. You can cd in and use plain git.
  • Not a hosted VPS product. It runs on your desktop by default. Remote compute uses SSH targets, self-hosted Orca servers, or cloud VMs you control.

It is for people who already write code, read diffs, and keep commits tidy — not a no-code tool. The marketing line is “IDEs were built for you. An ADE is built for you and your agents.”

How do you turn the CLI on?

One-line answer: Install the desktop app, register the CLI in Settings, then confirm orca can reach the runtime from a shell.

Start with the app. From Install:

  • Desktop builds for macOS, Windows, and Linux (AppImage, .deb, .rpm).
  • macOS Homebrew: brew install --cask stablyai/orca/orca. Stable upgrades: brew upgrade --cask orca.
  • Older builds: GitHub Releases.
  • The Linux desktop launcher is named orca-ide so it does not collide with the GNOME Orca screen reader. The scripting command documented below is orca.

On first launch Orca asks for home-directory access, offers to import ~/.claude, ~/.codex, and Ghostty settings if present, then drops you on an empty landing to add a first repo.

The CLI ships with the app. Overview says register it under Settings → General → Orca CLI. Reference says Settings → Experimental → CLI. Look for both in Settings; treat the live docs as source of truth if the label moves.

Agents can install the matching skill:

npx skills add https://github.com/stablyai/orca --skill orca-cli
# headless / no Settings UI:
orca skills install --skill orca-cli

orca skills install / update shell out to the same npx skills commands Settings uses. They do not contact the Orca runtime.

What are the first verify commands?

One-line answer: command -v orca for PATH, then orca status --json for the runtime. If the app is not running, orca open --json and status again.

command -v orca
orca status --json

If Orca is not already running:

orca open --json
orca status --json

Use --json when another tool will parse the result. Human-readable output is for a quick terminal check. orca talks to a running runtime. If the app (or orca serve) is down, a failed status is expected.

After a repo is added you can continue with the documented worktree checks:

orca worktree ps --json
orca worktree current --json

Docs point next to “Your first 3-agent session.” This article stops at install, registration, and status. Creating worktrees and picking agents is the next piece in the series.

Sources