We built Cyberdesk – an open source API for building computer agents that can control a full virtual desktop: click, type, scroll, and more.
If you're working on an agent that needs to operate real apps (desktop apps, web apps in a browser, etc.), this saves you the headache of setting up and scaling VMs. Just call our API to start a desktop and send actions.
Our entire infrastructure, frontend, backend, and SDKs are open source.
GitHub: https://github.com/cyberdesk-hq/cyberdesk
Docs: https://docs.cyberdesk.io
Demo: https://cyberdesk.io
Cyberdesk is ideal if you’re:
- Building screenshot based computer use agents
- Tired of duct-taping Docker, VNC, and cloud infra together
- Curious about computer use
We’re early and would love feedback — whether it's UX/API design, missing features, or performance.
Happy to answer anything. Thanks!
