Orca
Stably AI · Open-Source Environment for Running Coding Agents in Parallel
Orca is a free, MIT-licensed agent development environment (ADE) from Stably AI for developers who run CLI coding agents like Claude Code, Codex, and OpenCode side by side, each in its own git worktree with a terminal or chat view, a browser, and diff review. It does not supply a model — you bring your own agents and subscriptions.
Best for
Developers who already use one or more CLI coding agents and want to run and review many of them in parallel without branch conflicts
Not ideal for
Beginners or no-code users, anyone looking for a coding agent with its own model, or teams that need managed cloud execution
Who it's for
Professional developers running multiple terminal coding agents at once who want isolation, review, and monitoring in one app
Capabilities
- Runs each task in its own git worktree with a dedicated agent session and browser tab
- Optional structured chat view for Claude, Codex, Grok, OpenCode, Pi, OMP, and Antigravity, with model pickers, queued or steered follow-ups and, for some agents, plan mode and response forking
- Works with any CLI agent, with built-in integrations for Claude Code, Codex, and Cursor CLI and auto-setup for Gemini, OpenCode, GitHub Copilot CLI, Antigravity, Cline, Goose, and others
- Diff viewer with inline annotations on AI changes that can be sent back to the agent
- GitHub, GitLab, Bitbucket, Linear, and Jira integration for issues, pull or merge requests, checks, and committing or pushing work
- Per-worktree Chromium browser with Design Mode for inspecting HTML and CSS
- Remote SSH worktrees, always-on Orca servers you pair with from a laptop or phone, and experimental per-workspace cloud VMs on your own provider account
- CLI with scripting, scheduled automations, coordinator agents that start workers, worktree checkpoints, and computer use
- Usage and rate-limit tracking for Claude Code, Codex, Gemini, Cursor, and other plans, with account switching
- Plugin directory for themes, syntax highlighting, and custom editors
- iOS and Android companion apps to monitor agents and reply when they need input
Limitations
- Provides no model of its own — you need existing agent subscriptions or API keys (for example Claude, Codex, or OpenCode)
- Aimed at professional developers; assumes familiarity with git worktrees and terminal agents
- Mobile apps are in beta and act only as a remote control for a running, paired desktop or Orca server
- New terminal-agent launches pre-fill each CLI's permission-bypass flag (for example --dangerously-skip-permissions for Claude Code); choose Manual (or Auto for Claude and Codex) under Settings → Agents if you want approvals
- Plugins that run code get full access to your computer once you consent to install them
- Collects anonymous usage telemetry by default (opt-out available; prompts and code are not sent)
- No Orca-hosted compute — remote runs use SSH hosts, your own Orca servers, or experimental cloud VM recipes on a provider account you pay for
- Ships updates almost daily, so settings and features change often, and some (cloud VMs, WSL hosts, Chrome extensions) are still experimental
Use cases
- Running Claude Code and Codex on the same issue in separate worktrees and comparing their diffs
- Working several Linear or GitHub issues in parallel, each with its own branch and agent
- Annotating an agent's diff and sending review comments back for another pass
- Checking UI changes in the per-worktree browser with Design Mode before committing
- Running agents on an always-on home server or remote build machine and answering their questions from a phone
Orca pricing
Open source
Free
- MIT license
- Desktop apps for macOS, Windows, and Linux
- iOS and Android companion apps (beta)
Note: Orca itself is free. Costs come from the coding agents and model subscriptions or API keys you run inside it, plus any SSH hosts or cloud VM providers you choose to use. An Enterprise page offers rollout guidance and compliance documentation by email; it lists no paid plan. Verified on 2026-11-10.
Orca vs. competitors
Orca vs. Google Antigravity
Antigravity 2.0 also supervises parallel agents, but runs Google's own agent harness with Gemini and select other models; Orca is open source and agent-agnostic, orchestrating third-party CLI agents you already use.
Orca vs. Cursor
Cursor is an AI editor with its own agent and cloud background agents; Orca has no model of its own and focuses on running many CLI agents in separate worktrees, on your machine or on servers you control.
Orca vs. Claude Code
Claude Code is one agent working in your repo; Orca can run several Claude Code sessions, or a mix of different agents, side by side and compare their results.
Frequently asked questions
What is Orca?
Orca is a desktop agent development environment from Stably AI for running multiple AI coding agents side by side. Each task gets its own git worktree, agent terminal, and browser tab, with tools for reviewing diffs and shipping changes.
Is Orca free and open source?
Yes. Orca is free and released under the MIT license, with the source on GitHub at stablyai/orca. You still pay for whichever coding agents or model subscriptions you run inside it.
Which coding agents work with Orca?
Orca works with any CLI agent. Claude Code, Codex, and Cursor CLI have built-in integrations, and agents such as Gemini, OpenCode, GitHub Copilot CLI, Antigravity, Aider, Goose, Cline, and Continue are supported through auto-setup.
Does Orca send my code to Stably?
According to Orca's telemetry docs, it collects only anonymous usage events and never sends prompts, responses, terminal contents, file paths, or repo names. Telemetry can be disabled in settings or with DO_NOT_TRACK=1.
Orca vs Claude Code: what is the difference?
Claude Code is a coding agent. Orca is an environment that runs coding agents, including Claude Code, in parallel isolated worktrees and adds diff review, a browser, and issue-tracker integration around them.
Can Orca run agents on a remote server or in the cloud?
Yes, on machines you control. You can run worktrees on SSH hosts, keep an always-on Orca server (the desktop app or orca serve) that your laptop and phone connect to, or, as an experimental option, boot a per-workspace cloud VM or container from a recipe in your repo. Orca does not sell hosted compute, so cloud VMs run on and are billed by your own provider account.
Integrations & fit
Alternatives to consider
Developers who want to supervise several autonomous coding agents in parallel from a dedicated desktop app, with a free Gemini tier and Claude models on paid Google AI plans
FreemiumAnysphere
VS Code developers who want multi-file AI editing, autonomous cloud agents, and model flexibility — not a plugin layered on top of an existing editor
FreemiumAnthropic
Developers and teams who want deep codebase reasoning, multi-file changes, and persistent project instructions in a terminal-friendly or IDE-native workflow
Paid
Paperclip Labs, Inc.
Technical founders, builders, and small teams already running several coding or general-purpose agents who want budgets, approvals, and an audit trail around them
Free
Esoteric Labs
Developers who already run several Claude Code or Codex sessions on one project and want them organized into a persistent, addressable team they can attach to and steer from the terminal
Free
JetBrains
Engineering organizations on JetBrains IDEs whose developers use several coding agents and that want parallel agent sessions in the IDE, shared cloud tasks and automations, and one place to set agent policies, credit limits, and BYOK providers
Freemium
T3 Tools
Developers who already pay for one or more coding agents (Claude Code, Codex, Cursor, Grok Build, OpenCode) and want a single GUI to run them in parallel, ship pull requests, and keep working from a phone or another computer
FreeAbout Orca
Orca sits one layer above individual coding agents. Instead of choosing between Claude Code, Codex, Cursor CLI, Gemini, OpenCode, or dozens of other terminal agents, you launch them inside Orca, and every task gets its own git worktree, agent session, and browser tab so parallel work never collides on a branch. Each session can run in a terminal or, for Claude, Codex, Grok, OpenCode, Pi, OMP, and Antigravity, in a structured chat view with model pickers, queued follow-ups and, for some agents, plan mode and response forking. Around that core are the tools for supervising many runs at once: GPU-rendered terminals with unlimited splits, a Monaco editor, a diff viewer where you annotate AI changes and send the comments back to the agent, GitHub, GitLab, Bitbucket, Linear, and Jira integration for issues and pull requests, a per-worktree Chromium browser with a Design Mode for inspecting HTML and CSS, a status-bar readout of Claude, Codex, and other plan limits with account switching, and a plugin directory for themes, languages, and custom editors. Agents can run on your desktop, on SSH hosts, on an always-on Orca server (the desktop app or orca serve) that your laptop and phone pair with, or, as an experimental option, on per-workspace cloud VMs or containers defined by a recipe in the repo and billed by your own provider; Orca does not sell hosted compute. A CLI adds scripting, scheduled automations, orchestration of worker agents, and computer use, and iOS and Android companion apps (in beta) act as a remote control for the desktop or server that owns the sessions. Orca is free and open source, runs on macOS, Windows, and Linux, ships new releases almost daily, and its telemetry excludes prompts, code, and repo names. The tradeoffs are that it adds nothing to model quality, since results depend on the agents and subscriptions you already pay for, and that it starts terminal agents with their permission-bypass flags by default, on the basis that worktrees are disposable; switch to Manual approvals (or Auto for Claude and Codex) in settings if you want agents to ask first.
Updates from Orca
Orca 1.4.224 made structured native chat the default for Chat UI users, opened OpenCode, Pi, and OMP as structured chats, and added managed Orca runtimes on SSH hosts and inline charts and diagrams in chat replies.
Orca 1.4.245 added installable plugins for themes, syntax highlighting, and custom editors, with consent required for plugins that run code, and let agents use Mac apps in the background while you keep working.
Are you the founder? Claim this listing →