Comparisons

boxes.dev vs Orca

Orca is a free, open-source agent IDE for running many CLI agents in Git worktrees on your computer or hosts you run. boxes.dev runs the machines for you and gives each task a full, isolated copy of your development environment.

The short answer

Orca is a free, MIT-licensed agent IDE for running Claude Code, Codex, and more than 30 other CLI agents in Git worktrees on your computer, an SSH host, or an Orca server you run, with iOS and Android companions. boxes.dev runs Claude Code and Codex on full, isolated copies of a cloud machine where your app already runs. Choose Orca for a free IDE with many agents and stacked pull requests. Choose boxes.dev when you want the agent to test its work in your full running app, like an engineer would, without running hosts, and with Slack and Linear starts.

Choose boxes.dev if

  • You want the agent to test its changes in your running app, with its real databases, services, and data, the way an engineer would.
  • You don't want to provision, secure, and keep hosts online for your agents.
  • You want each task on its own machine, with its own databases and ports, that sleeps when idle and resumes with its dev servers still running.
  • You want work to start from Slack mentions, assigned Linear issues, or an API call.
  • You want to share the app a task is running at a public link, or give an OAuth provider or webhook one stable Project URL.

Choose Orca if

  • You want free, MIT-licensed software you can change and fork.
  • You use agents beyond Claude Code and Codex, and Orca has more than 30 ready to launch.
  • You want to create GitHub stacked pull requests and turn on auto-merge from the same app.
  • You want agents to control an iOS Simulator through the Orca CLI.

At a glance

As of September 2026.

boxes.devOrca
AgentsClaude Code and Codex, each with your own plan or an API key, in a graphical view or the agent's own terminal UIClaude Code, Codex, Cursor CLI, and more than 30 other CLI agents ready to launch, or any other CLI command, each signed in on the host
Who runs the computerboxes.dev runs a cloud devbox for each task: a full, isolated copy of your Template box that can hold several threadsYou do: your computer, an SSH host, an Orca server, or an experimental cloud recipe in your own provider account
Parallel tasksEach devbox is its own machine, with its own databases, services, and portsGit worktrees, each with its own branch and files, on the same host
Environment setupA snapshot of your Template box, a cloud machine that a setup agent prepares with your app's tools, databases, services, and data, kept current by an optional maintenance scriptThe host as you set it up; commands that run when a worktree is created, and ignored files copied or shared from the main checkout
When you close your laptopDevboxes keep working, then sleep when idle, keeping files and memory, so dev servers, databases, and terminals resume when they wakeAgents on an SSH host or Orca server keep working while that host stays awake and connected
Working in the machineTerminal tabs, a file editor, Git history and diff review, a shared browser with comments on page elements, a process monitor, and SSHTerminals, a code editor, diff review, a browser with DevTools and Design Mode, and simulator controls
Reaching the running appA browser on the devbox that you and the agent share, ports forwarded to the same port on your laptop, stable .localhost addresses, public links, and Project URLsA browser for each worktree that agents can drive, with traffic routed through a remote host; an SSH host's listening ports forwarded to your laptop
AppsmacOS desktop app, iOS and Android apps with a terminal, the dvb CLI, SSH, and VS Code or JetBrains through dvb ideDesktop app, a browser client for Orca servers, iOS and Android companions in beta, the orca CLI, and VS Code for SSH worktrees
Starting workThe apps, dvb claude or dvb codex from your terminal, @boxes.dev in Slack, assigned Linear issues, and automations on a schedule or from an API callThe apps, the orca CLI, worktrees created from GitHub or Linear issues, and automations on a schedule on a host you choose
Pull requestsComments on changed lines go to the agent as one message; actions to fix checks, address review feedback, and merge; and pull requests that fix failed checks and bot review feedback on their ownLine notes on diffs go to the agent as one batch; an action that sends failed checks to an agent; checks, review comments, auto-merge, and GitHub stacked pull requests
Shared setupEnvironment files with project, team, and personal values; startup and maintenance scripts; MCP servers and agent plugins on every devbox; AWS, Tailscale, and a static egress IP; Team TemplatesA checked-in orca.yaml, plus each host's own agent logins, MCP servers, and settings
Working with your teamRead-only thread links, team-wide thread reading, and project browsing for teammates; @boxes.dev in Slack, where teammates you allow can start work and send follow-ups that run with your access; and commits linked to threadsSeveral clients paired to one Orca server, each with access you can revoke, and public links for HTML and Markdown files
Agents working togetherAgents start agents on new or existing devboxes, message each other, read earlier threads, and share files in a project folderExperimental orchestration, where a coordinator dispatches tasks to supervised workers across worktrees and hosts
PricingFree trial with 10 box-hours; Starter $19 and Pro $99 per user each month, plus your Claude or ChatGPT plan or an API keyFree and MIT-licensed; you pay for your hosts and agent plans

Who runs the computer

Orca runs agents on computers you provide, and boxes.dev runs the computers for you.

Orca runs worktrees on your computer, on an SSH host, or on a Remote Orca Server, where a desktop or headless orca serve host keeps the projects, terminals, and agent sessions, and your laptop, a browser, or your phone connects to it. You install and sign in to each agent's CLI on that host, keep it on a private network such as Tailscale, and keep it secure, updated, and awake. An experimental Cloud VM mode runs create, suspend, resume, and destroy scripts from your repository against your own cloud provider.

boxes.dev first sets up your full development environment on a cloud machine called the Template box, with the tools, databases, services, and data your app needs, so an agent can run the app and test its changes end to end. A setup agent can port the environment from your laptop, including the files, local data, agent settings, and skills you choose to upload, or build it from a GitHub repository. boxes.dev then saves the Template box as a snapshot, and every task runs on its own devbox, a full, isolated copy of that machine. A maintenance script can keep the Template box current as your code changes, running on a schedule or when files such as a lockfile change, and each successful run saves a new snapshot. Team Templates let an admin prepare one starting environment for everyone on the team.

You'd never let an engineer ship code they hadn't run, and a full environment holds the agent to the same standard. On a boxes.dev devbox, the agent runs your whole app, with its databases and services, and tests its changes in the browser as it builds. It reproduces bugs in the real app to find the root cause instead of guessing from the code, so it writes better code and makes fewer mistakes. It can also take screenshots and record video for you to review, so you can hand it most repetitive manual testing.

Worktrees on one host, or a machine for each task

Orca separates parallel tasks with Git worktrees, and boxes.dev gives each task its own machine.

A new Orca worktree is a clean checkout on its own branch. Commands can run when it's created, and gitignored files such as .env or node_modules can be copied or shared from the main checkout. The worktrees share the host's CPU, memory, services, databases, and ports, so tasks that each need their own database or a dev server on the same port need more setup, such as a Cloud VM recipe.

Each boxes.dev devbox is a separate machine with its own copy of your databases, services, and seeded data, so parallel tasks don't collide.

Reaching the running app

Both let you and your agents use the running app in a browser, and boxes.dev also shares it at public addresses.

Each Orca worktree has its own Chromium browser that agents can drive through the Orca CLI. For a workspace on an Orca server or SSH host, the browser's traffic can go through the host, or the server can stream the page. For SSH worktrees, a Ports tab lists the host's listening ports and forwards them to your laptop in one click. Orca can also publish self-contained HTML and Markdown files at public links.

Each boxes.dev devbox runs a browser that you and the agent share, so you can watch the agent test the app, click through it yourself, or comment on page elements to show the agent what to change. Ports forwards the app to the same localhost port on your laptop and gives each devbox stable .localhost addresses. A public link shares a devbox's running app with anyone, and a Project URL gives your project one stable public address that you can point at any devbox, which suits OAuth callbacks and webhooks.

When you close your laptop

Both keep working when the work runs on another machine, and boxes.dev also sleeps machines that nobody is using.

An SSH host or Orca server keeps agents running while your laptop sleeps, as long as the host stays awake and connected, and the iOS and Android companions and scheduled automations reach it there.

A boxes.dev devbox keeps working while your laptop is closed, then sleeps when idle and keeps its memory, so dev servers and databases resume where they left off. Schedules, Slack mentions, and assigned Linear issues start work on devboxes, and the iOS and Android apps can reply to agents, browse files, and open a shell.

What comes with every boxes.dev devbox

The boxes.dev desktop app opens the same set of tools for whichever devbox you select:

  • Terminal, with tabs that keep running after you close the app.
  • Files, a file browser and editor with search across the workspace.
  • Git/Review, for branch history and diffs, where comments on changed lines go to the agent together as one message.
  • Browser, the browser on the devbox that you and the agent share.
  • Ports, for the forwarding, .localhost addresses, and public links described above.
  • Activity, which shows the processes using the machine's CPU and memory.

In Slack, later @boxes.dev mentions from you or teammates you allow send follow-ups to the same conversation, and those teammates can start work in your project with @boxes.dev @YourName, which runs with your access. AWS, Tailscale, and the MCP servers and agent plugins you add for the project reach every devbox. You can branch a conversation onto a new devbox that starts with the current files, including uncommitted work.

Where Orca shines

  • Free and open source. Orca is MIT-licensed, so you can read, change, and fork the whole app.
  • More than 30 agents. Claude Code, Codex, Cursor CLI, and dozens more launch in one click, with status, usage tracking, and account switching for the most deeply integrated ones.
  • Stacked pull requests and auto-merge. Orca can create GitHub stacked pull requests, show where a pull request sits in its stack, and turn on GitHub auto-merge from the pull request view.
  • Scriptable, including simulators. The orca CLI drives worktrees, terminals, the browser, an iOS Simulator, and automations, so agents can operate Orca too.

Pricing

As of September 2026.

boxes.devOrca
PriceTrial free; Starter $19 and Pro $99 per user each monthFree, with no paid plans published
Model usageYour Claude or ChatGPT plan, or an API keyEach agent's own plan or API key
Cloud time included10 box-hours on the Trial, 40 per user on Starter, 250 on ProNone; you provide the computers
Cloud machines awake at once2 on the Trial, 4 per user on Starter, 10 on ProAs many hosts as you run; worktrees on one host share it
Extra cloud time$0.75 per box-hour on Starter, $0.60 on ProYour hosting provider's rates

A box-hour is one default devbox awake for an hour, and sleeping devboxes don't use any. On Starter, 40 box-hours works out to about two hours of devbox time each working day, and Pro's 250 to about twelve, spread across as many as ten devboxes at once. Your Claude or ChatGPT plan covers the agents' usage. See Plans, seats, and box-hours.

Orca is free and MIT-licensed, and its site lists no paid plans. Its Enterprise page offers rollout guidance and support through its team. You pay your agent providers and any hosts or cloud accounts your worktrees run on. See Orca's enterprise page and ways to run Orca.

Using both

Orca and boxes.dev can work on the same repositories and pull requests. You can keep Orca for the agents and hosts you already run with it, and use boxes.dev when the agent should test its work in its own copy of your full running app, or when work should start from Slack, Linear, or an API call.

Frequently asked questions

Is Orca free?

Yes. Orca is MIT-licensed and publishes no paid plans, and you pay for your agents' plans and any hosts you run it on.

How does a Remote Orca Server compare with boxes.dev?

A Remote Orca Server keeps agents running on a host you control, but you provision, secure, update, and pay for it, and its worktrees share its resources. boxes.dev gives each task its own copy of your Template box and sleeps idle machines.

Can I use my Claude and ChatGPT plans with both?

Yes. Orca launches Claude Code and Codex with the logins on your host, and boxes.dev runs them with your Claude or ChatGPT plan or an API key.

Sources

Product names are trademarks of their respective owners.

Close your laptop again.

Develop in the cloud: one computer per agent, running your full app, steerable from every device you own.

Also oniOS·Android·CLI