Orca

Orca

The AI Orchestrator for 100x builders. Orca is designed to empower developers to run Claude Code, Codex, or OpenCode side-by-side across multiple repositories, with each agent operating in its own worktree and managed from a single, cohesive interface. It’s built to be platform-agnostic, supporting macOS, Windows, and Linux. With Orca, you get a centralized control hub for your AI-assisted development workflow—without the friction of juggling countless terminals, stash operations, or context switches.

Install and try Orca today

Introduction: Why Orca Exists

In modern software development, teams often rely on a mix of AI agents to accelerate coding, debugging, and design tasks. Claude Code, Codex, OpenCode, and other agents each bring unique strengths—and each often lives in its own ecosystem, requiring separate workspaces or contexts. Orca reimagines this landscape by acting as an AI orchestrator that federates these agents under a single, worktree-native umbrella. The result is a workflow where:

• You can run multiple AI agents in parallel, side-by-side, across different repositories.
• Each agent has its own worktree so changes, diffs, and progress are isolated and easy to manage.
• You gain built-in source control that lets you review AI-generated diffs, make quick edits, and commit without leaving Orca.
• GitHub integration links PRs, issues, and Actions checks to the corresponding worktrees automatically, keeping your development lifecycle aligned with your repository’s workflow.
• SSH support lets you connect to remote machines and run agents directly from Orca, expanding your reach beyond a single workstation.
• Real-time notifications ensure you know when an agent finishes a task, needs attention, or has generated results worthy of a quick review.

Orca is designed for developers who want to focus on building, not on managing their AI tooling. It’s the kind of tool you can rely on every day to keep your AI-assisted workflow fast, transparent, and auditable.

Key Concepts: How Orca Keeps AI in Sync with Your Code

• Worktree-native architecture: Each feature and every AI agent operates within its own worktree. This ensures clean separation of experiments, reduces context-switching costs, and eliminates the need for constant stashing or branch juggling.
• Multi-agent terminals: Orca presents multiple AI agents in tabs and panes. You can see at a glance which agents are active, what they’re working on, and what outcomes have been produced. This visual organization is designed to remove guesswork from parallel AI collaboration.
• Built-in source control: Diffs generated by AI are shown inside Orca. You can edit, refine, and commit changes without leaving the interface, turning AI-assisted outcomes into tangible, versioned code in a single flow.
• GitHub integration: Contact with GitHub is woven into the worktree structure. PRs, issues, and Actions checks become linked artifacts that automatically reflect the state and progress of each AI-driven worktree.
• SSH support: If your development environment spans multiple machines, Orca lets you operate agents on remote hosts as if they were local. SSH-enabled orchestration empowers distributed AI workflows with minimal friction.
• Notifications: Every action, completion, or required attention is surfaced through timely notifications. You can mark threads as unread to return to them later without losing track of your AI-assisted tasks.

Supported Agents: A Rich Ecosystem of AI Tools

Orca is designed to be agent-agnostic. It supports any CLI agent, enabling you to mix and match to suit your project, token budget, or preferred workflow. The list below highlights a broad spectrum of popular AI engines and interfaces you can pair with Orca:

• Claude Code
• Codex
• Gemini
• Pi
• Hermes Agent
• OpenCode
• Goose
• Amp
• Auggie
• Charm
• Cline
• Codebuff
• Continue
• Cursor
• Droid
• GitHub Copilot
• Kilocode
• Kimi
• Kiro
• Mistral Vibe
• Qwen Code
• Rovo Dev

These names are presented with the idea that Orca doesn’t lock you to a single provider. Instead, you bring your own subscriptions and tokens, and Orca coordinates the tools you rely on across your codebase. The result is a single pane of glass for your AI-assisted development, regardless of which agent you prefer for a given task.

Features: What Makes Orca Stand Out

• No login required: You bring your own Claude Code or Codex subscription. Orca doesn’t require its own centralized login for agents; it orchestrates the tools you already possess.
• Worktree-native operations: Every feature gets its own worktree. You don’t need to stash changes or juggle multiple branches to experiment with AI-driven improvements.
• Multi-agent terminals in one view: Run several AI agents at once in a single window, with clear visual separation. It’s easy to see which agent is active and what stage each task is in.
• Built-in diffs and quick edits: Review AI-generated changes directly inside Orca, tweak as needed, and commit—without leaving the interface.
• GitHub integration by default: PRs, issues, and Actions checks are connected to each worktree, maintaining alignment with your project’s lifecycle.
• SSH support across machines: Run agents on remote hosts and manage them from your main workstation.
• Real-time notifications: Stay informed about task completions and outstanding prompts. Mark threads as unread when you want to revisit a conversation later.

Install: Getting Orca Up and Running

Mac, Linux, Windows

• Download from onOrca.dev
• Or via GitHub Releases page
• This path is ideal if you prefer a static binary or want a quick bootstrap.
• Package manager options:
• macOS (Homebrew)
  • Command: brew install --cask stablyai/orca/orca
• Arch Linux (AUR)
  • Precompiled binary: yay -S stably-orca-bin
  • Build from GitHub source: yay -S stably-orca-git

The install path emphasizes flexibility: you can grab a prebuilt artifact, or pull Orca through your preferred package manager and keep it up to date with system updates. Regardless of the route you choose, Orca is designed to integrate into your existing development environment with minimal friction.

New: Annotate AI Diff

One of the thoughtful enhancements Orca brings is the ability to annotate AI-generated diffs directly. Instead of copying line numbers into a separate comment or retyping suggestions, you can comment on any line within an AI-generated diff. Your feedback returns to the agent, which can revise the changes accordingly. This tight feedback loop helps you refine AI outputs without juggling context.

• Comment on specific lines in AI-generated diffs to guide the agent’s revisions
• Reduce context-switching: keep your review loop inside Orca
• See revised diffs in the same interface for rapid iteration

Illustration: Orca Annotate AI Diff

New: Hot Swap Codex Accounts

For teams or individuals who run multiple Codex accounts to optimize token usage or access different capabilities, Orca adds a convenient hot-swap capability. Switch between Codex accounts in one click—no re-login, no reconfig files, just select the desired account and continue building.

• Quick account switching across Codex instances
• No login prompts or configuration edits required
• Keeps token usage flexible and cost-efficient

Illustration: Orca Codex Account Switcher

New: Per Worktree Browser & Design Mode

Orca introduces a built-in browser inside your worktree, allowing you to preview your application as you build. This is followed by Design Mode, a powerful feature that lets you click any UI element in the preview and drop it directly into your AI chat as context. No need for screenshots, no manual selector copying—the agent gets precise, actionable context.

• See your app in real time within the same workspace
• Click any UI element to drop it into chat as design context
• Eliminate tedious screenshot and selector-copying workflows

Illustration: Orca Design Mode

New: Introducing the Orca CLI

Orca’s capabilities extend to the terminal with the Orca CLI, bringing agent orchestration from the command line into your IDE workflow. The CLI can add projects to your IDE, spin up worktrees, and update the current worktree’s comment with meaningful progress checkpoints—directly from the terminal. It ships with the Orca IDE, which you can install under Settings.

• Orchestrate AI agents from the terminal
• Manage worktrees and progress checkpoints programmatically
• Integrates with the built-in Orca IDE for a cohesive experience

Example command: npx skills add https://github.com/stablyai/orca --skill orca-cli

Community & Support: Connect, Share, and Grow

• Discord: Join the community on Discord
• Twitter / X: Follow @orca_build for updates and announcements
• Feedback & Ideas: If something is missing, request a new feature
• Show Support: Star the repository to follow nightly ships and updates

Developing: Contributing and Local Development

If you’re interested in contributing or running Orca locally, the project provides a clear path. See the CONTRIBUTING.md guidance to set up a development environment, run tests, and participate in the ongoing evolution of Orca.

• Access to the CONTRIBUTING.md: .github/CONTRIBUTING.md
• Guidance for setting up a local development environment
• Information on contributing, issue reporting, and release processes

A Seamless Experience Across Your Toolchain

Orca isn’t about replacing the tools you love; it’s about orchestrating them so you can move faster, with more visibility and less friction. The worktree-centric approach ensures that each AI-driven experiment stays isolated and auditable. The multi-agent terminal environment makes it simple to compare approaches—running Claude Code on one worktree while Codex runs on another—and then unify results by integrating them back into your codebase with built-in diffs and one-click commits.

In practice, Orca helps teams manage complicated AI-assisted workflows with:

• Transparent progress tracking: You can clearly see what each agent is doing, what has completed, and what still needs attention.
• Coherence between AI outputs and code: AI suggestions are not black boxes; you can review and refine them in the same interface that manages your code.
• A familiar GitHub-connected lifecycle: PRs and checks are linked to worktrees so your AI-assisted changes travel through the same gates as your manual edits.
• The convenience of SSH and remote work: If your development environment sits across multiple machines or cloud instances, Orca gives you a direct line to run agents remotely without breaking the flow.

Visual Tour: A Glimpse at Orca in Action

• The Orca Screenshot GIF offers a quick sense of how files can be dragged into the workspace, and how the interface manages multiple agents and worktrees.
• The Annotate AI Diff GIF demonstrates the new in-diff commentary capability, highlighting how feedback travels back to the agent for revision.
• The Codex Account Switcher GIF shows how you can hop between Codex accounts without re-authenticating.
• The Design Mode GIF reveals the intuitive process of selecting a UI element and transferring that context into the AI chat for precise guidance.
• The Annotate AI Diff and Design Mode features together illustrate a modern, feedback-rich loop between human review and AI generation—an essential pattern for reliable AI-assisted development.

How to Get the Most from Orca

• Start with a clear separation of concerns: assign each feature or task to its own worktree. This keeps experiments isolated and reduces cross-talk between AI tasks.
• Leverage the multi-agent view to compare strategies: run different agents on similar tasks to gauge strengths and identify best-fit approaches for your project.
• Use built-in diffs as the source of truth: edit and commit directly within Orca to maintain an auditable history of AI-assisted changes.
• Integrate with GitHub early: link worktrees to PRs and checks to ensure your AI-driven improvements flow smoothly into your mainline codebase.
• Explore the Design Mode for UI-centric apps: if you’re building front-end interfaces or design-driven features, this mode provides contextual AI guidance without text-only prompts.

A Note on Accessibility and Collaboration

Orca is designed to be approachable for both solo developers and teams. By consolidating AI orchestration into a single interface, it reduces cognitive overhead and accelerates onboarding for new team members who need to understand how AI agents contribute to the project. The design emphasizes clear visual cues, straightforward navigation between worktrees, and concise feedback loops so everyone can participate in AI-driven development without getting lost in the maze of tools.

Conclusion: A Central Nervous System for AI-Enhanced Development

Orca is more than a collection of features; it is a strategic approach to integrating AI into the everyday flow of software creation. The combination of worktree-native architecture, multi-agent collaboration, built-in version control, first-class GitHub integration, and remote-agent capabilities means you can compose complex AI-assisted workflows with confidence. The recent feature additions—annotating AI diffs, hot-swapping Codex accounts, per-worktree browsing and Design Mode, and the Orca CLI—underscore a commitment to speed, clarity, and control.

If you’re seeking a way to bring together Claude Code, Codex, OpenCode, and other AI agents under one roof, while preserving a clean, auditable, and scalable workflow, Orca presents a compelling path forward. It is designed to adapt to your preferred agents, your project structure, and your preferred OS, all while delivering a unified experience that keeps AI assistance aligned with your real-world coding goals.

Install, try, and contribute today. The future of AI-enhanced development is here, and Orca is ready to orchestrate it with you.

Author’s note: The content above is drawn from the Orca project documentation and release notes embedded in the input resource. The descriptions and feature highlights reflect the capabilities outlined in the provided content, including the new annotations, Codex account switching, per-worktree browser and design mode, and the Orca CLI. For the latest updates and more detailed guidance, please refer to the official Orca repository and documentation.