Short answer
Orca and Orchestrator.inc (formerly Agent Orchestrator) are both free, open source, cross-platform apps for running coding agents in parallel worktrees, with mobile companions. The difference is the shape: Orca is an agent IDE you work inside, with rich PR tooling across many Git forges, while Orchestrator is a supervisor that sits beside your editor and sends failing CI checks and review comments to the owning agent without a click. Choose Orca if you want an IDE-style app, Design Mode, or Bitbucket, Azure DevOps or Gitea. Choose Orchestrator if you want a board of sessions and the PR follow-up handled for you.
At a glance
| Orchestrator.inc | Orca | |
|---|---|---|
| Shape | Supervisor: desktop app, local daemon and ao CLI | Agent IDE with terminals, PR views and a browser per worktree |
| Platforms | macOS, Windows x64, Linux x64, plus iOS and Android companion | macOS, Windows and Linux, plus iOS and Android companions; experimental WSL hosts |
| License and price | Apache-2.0, free locally with no account; optional Cloud plans from $20/month | MIT, free; sales-led Enterprise tier |
| Coding agents | 38 harnesses, including Claude Code, Codex, Cursor, OpenCode and Gemini CLI | Claude Code, Codex, Cursor, Copilot, OpenCode, Amp and other CLI agents |
| Isolation | Git worktree and branch per session; multi-repo workspaces | Git worktrees, with optional per-worktree sandbox, VM or Docker environments |
| Remote | Host on your own server or VM, several hosts in one desktop app, linked by account or tunnel; mobile pairs over LAN or a managed tunnel | SSH hosts, a self-hosted server and a web client |
| Overview | Kanban board with lanes derived from session, PR, CI and review facts | No native sessions board as of October 2026; browses GitHub Projects, Linear and Jira boards |
| CI and review | Failing checks, review comments and merge conflicts sent to the owning session automatically | Failed checks shown inline with logs; "Fix broken checks" and "Resolve with AI" buttons |
| Forges | GitHub and GitLab, including self-managed GitLab | GitHub, GitLab, Bitbucket, Azure DevOps and Gitea |
| Planner agent | Project orchestrator agent that plans, spawns and briefs workers | Experimental orchestration CLI driven by an agent |
The problem in practice
A PR has one failing check and one review comment, and it needs to merge before lunch. Both tools show you the failure. The question is who carries it to the agent. If you have six sessions open, every hand-off you do yourself is a context switch: notice the red chip, open the right worktree, pass the failure along, read the comment, explain it to the agent, check again.
How each tool handles it
In Orca
- Open the worktree. The PR view shows the failed check with inline logs.
- Click "Fix broken checks". The agent receives the failed check names and works in the terminal.
- Read the review comment in the inline review view and tell the agent what to change.
- Push, then merge, or let auto-merge finish once checks are green.
In Orchestrator.inc
- The owning session already has its PR, discovered when the agent pushed.
- The check fails. Orchestrator sends the failing check names, links and the end of each failed job's log to that session, and the agent fixes and pushes.
- The review comment arrives as focused feedback in the same session, and the agent addresses it.
- The card reaches Ready to merge. You read the diff and merge. Merge is always your explicit action.
Orca makes each hand-off one click and can auto-merge at the end. Orchestrator sends the failure and the comment without a click and leaves the merge to you.
Where Orca is a good fit
- You want an IDE-style workspace. Terminals with splits, PR views and a Chromium window per worktree with Design Mode for visual UI work. Orchestrator has no code editor.
- Your code lives beyond GitHub and GitLab. Orca works with Bitbucket, Azure DevOps and Gitea, and supports stacked PRs and auto-merge.
- You want plain SSH, a web client or sandboxes. Orca reaches remote machines over SSH, offers a web client, and can give each worktree its own sandbox, VM or Docker environment.
Where Orchestrator fits better
- Workflows where CI and review feedback should reach the agent on its own. Orchestrator polls GitHub and GitLab with ETags, with no webhooks or GitHub App, and sends failing checks, requested changes and merge conflicts to the owning session, de-duplicated.
- People who want a board of sessions. Cards sit in Working, Needs you, In review and Ready to merge based on session, PR, CI and review facts. Nobody drags them.
- Larger, planned pieces of work. A project orchestrator agent breaks a plan into tasks, spawns and briefs workers, and follows their progress. A reviewer agent from another vendor, such as Codex reviewing a Claude Code PR, can run on each PR.
- Agents on your own servers. Orchestrator can run its host on a server or cloud VM, and you start and supervise sessions there from your desktop, with several hosts in one app. A remote session's agent browser shows in your Browser panel.
- People who prefer their own editor. Orchestrator sits beside it. Its Files panel shows uncommitted, staged, per-commit and whole-branch diffs, and your inline line comments go to the agent.
One tradeoff on Orchestrator's side: local orchestration needs the machine on for automations and mobile access. The Stay Awake setting handles that, and mobile keeps working with a MacBook lid closed. To keep agents working with your laptop off, run them on a remote host or on Orchestrator Cloud.
Pricing
Running Orchestrator on your own machines is free and open source (Apache-2.0), with no account needed; you bring your own agent subscriptions or API keys, and it does not proxy model calls. Optional Orchestrator Cloud plans, for agents that keep working while your laptop is closed, are Starter at $20/month and Pro at $40/month, with Enterprise pricing on request. Orca is free under MIT, with a sales-led Enterprise tier, as of October 2026.
Who should pick which
Pick Orca if
- you use Bitbucket, Azure DevOps or Gitea, or rely on stacked PRs and auto-merge
- Design Mode in a real browser window matters to you
- you want to reach remote machines over plain SSH without installing a host on them, a web client, or per-worktree sandboxes
- you want one app to be where you write and review code
Pick Orchestrator if
- you want CI failures and review comments routed to the owning agent with no manual step
- you want a Kanban of sessions whose state is derived, not hand-set
- you want a planning agent and cross-vendor reviewer agents in the product
- you would rather keep your current editor and add a supervisor beside it
Try it
Install Orchestrator, spawn two workers on two issues, and let one PR fail a check to see the routing; the review loop guide explains each step.
Frequently Asked Questions
Is Orca open source?
Yes. As of October 2026 Orca is MIT licensed and describes itself as free and open source. Orchestrator.inc is Apache-2.0. Both are OSI-approved licenses.
Does Orca work on Windows and Linux?
Yes. Orca ships for macOS, Windows and Linux, with experimental WSL hosts. Orchestrator.inc also ships for all three, with a native ConPTY terminal host on Windows and AppImage, deb and rpm packages on Linux.
Does Orca fix failing CI automatically?
Orca shows failed checks with inline logs and offers a "Fix broken checks" button that hands the failed check names to an agent, so a person starts the fix. Orchestrator.inc sends the failing check names, links and the end of each failed job's log to the owning session as soon as it sees the failure.
Is there an Orca alternative with a Kanban board for sessions?
Orca does not have a native sessions board as of October 2026; it browses GitHub Projects, Linear and Jira boards instead. Orchestrator.inc has a Kanban whose lanes come from each session's PR, CI and review state. If that is what you are after, you can download Orchestrator.inc from orchestrator.inc and try it on one repo.