Pick your cloud
Choose the provider that fits each job, from quick sandboxes to infrastructure you already trust.
Heavy runs and throwaway sandboxes, on any cloud you pick. Your keys. Open source.
Big repos and long agent runs on cloud hardware. Keep light work on your laptop.
Try a repo, open a port, shut it down. Off your machine, still reachable from your devices.
Branch, commit, and push to GitHub like you would locally. Cloud machine, laptop habits.
Give each task its own workspace. Run several agents at once, then shut them down when the work is done.
Open your workspace from any device. No second computer to maintain, and you only pay for the compute you use.
Choose the provider that fits each job, from quick sandboxes to infrastructure you already trust.
Launch a workspace for one quick job or keep it available for ongoing work. You decide when it stops.
Securely reuse your model and environment credentials without pasting secrets into every workspace.
Open the same workspace in your browser or local client and continue exactly where you left off.
Stop active compute, preserve the workspace state, and resume when you are ready to keep going.
Expose any workspace port through a shareable URL for testing, feedback, or a quick demo.
Share a workspace with teammates when a task needs another set of eyes or a clean handoff.
Create and manage workspaces from scripts, internal tools, or the terminal you already use.
Hosted or self-hosted. No lock-in.
Or run it on your own infra
No. Use it for heavier runs, temporary sandboxes, or agents you want off your machine but still reachable.
Yes. Your Anthropic, OpenAI, or other keys. Encrypted, reused across workspaces, never marked up.
E2B, Daytona, Railway, AWS, Cloudflare, Ascii, Vercel, or exe.dev. Configure the ones you want and pick per workspace. No lock-in.
Yes. MIT-licensed, full stack on GitHub. One-click on Railway, or Docker anywhere.