Boxes.dev
Cloud VM for coding agents — run Codex and Claude Code threads in parallel, each on its own Linux box
| What is it | Cloud VM for coding agents — run Codex and Claude Code threads in parallel, each on its own Linux box |
|---|---|
| Pricing | Freemium — from $19/mo |
| Free tier | Yes |
| Platform | Web Application |
| API | Yes |
| Best for | running multiple coding agents in parallel on different tasks, letting agents work overnight on complex refactors or audits |
| Domain registered | 2026 |
Data updated Aug. 1, 2026
What does Boxes.dev do?
Boxes.dev is a cloud development environment built specifically for AI coding agents. Instead of running Codex or Claude Code on your local machine, Boxes.dev gives each agent thread its own full Linux VM in the cloud — with its own filesystem, services, ports, and database state. You can spin up multiple isolated "forks" from your main devbox, run agents in parallel, and leave them working even after you close your laptop. The tool works with the same Codex or Claude Code subscription you already use, and supports desktop (macOS), mobile (iOS, Android), web, and CLI access.
What makes Boxes.dev stand out is how it handles agent isolation and persistence. Each fork is a real computer, not just a git worktree — so agents can run the full app, take screenshots, test changes, and leave reviewable work behind. The desktop workbench shows a thread list with fork groups and live status, an integrated file viewer with diff mode, inline review comments, and persistent terminal sessions that survive disconnects. You can also set up automations: schedule weekly dependency audits, trigger threads from Slack or Linear, or have agents open PRs when work is done. Idle forks sleep automatically, so you only pay for compute time when agents are actually running.
Boxes.dev is most useful for developers who already use AI coding agents and want to run multiple tasks in parallel without bogging down their local machine. It's also handy for teams that need agents to keep working overnight or while someone is away. The free tier includes 10 box-hours, and paid plans start at $19 per user per month for 40 included hours. If you've ever wanted to kick off five agent tasks at once and come back to reviewable results in the morning, Boxes.dev makes that straightforward.
Key features
What makes it stand outWho is Boxes.dev for?
Who benefits most from this toolPricing
Free tier available — start without a credit cardTrial
- 10 free box-hours
- Main devbox + 2 awake forks
- Unlimited sleeping forks
- Bring your own Codex + Claude Code
- Desktop, mobile, web, CLI
Starter
- 40 included box-hours / mo
- Main devbox + 4 awake forks
- Unlimited sleeping forks
- Prepaid overage at $0.75 / box-hour
- Email support
Pro
- 250 included box-hours / mo
- Main devbox + 10 awake forks
- Unlimited sleeping forks
- Prepaid overage at $0.60 / box-hour
- Priority support
Teams
- Custom seat & box-hour pools
- SSO, SCIM, RBAC
- Audit logs and spend controls
- Dedicated resources & deployment
- Custom MSA, security review
Trust & presence
Gallery
Click any image to enlargeAlternatives in Developer Tools
AI coding assistant with autonomous agents — get real-time code suggestions, debugging help, and task automation.
Give AI coding agents safe, read-only access to your production database, logs, and analytics for better debugging.
Spawn coding agents that run infinitely in the cloud — describe what to build, and the agent ships real code autonomously.
Desktop IDE that lets you run Claude Code, Codex, Gemini, and other AI coding agents side by side in one workspace
Verdent coordinates multiple AI agents to tackle complex coding tasks in parallel. Sessions can run on their own while you step away, collision-free execution, clear insight, and a seamless flow that turns ideas into real, shippable code.
Multi-agent AI coding assistant that runs Claude, Codex, Gemini, and BlackBox agents in parallel for code tasks
Shared workspace where AI agents and humans collaborate in real-time on documents, tables, and tasks.
AI-native developer tools that help you write, debug, and maintain code using inference-time scaling and agentic systems