Rafael Viscarra, one of our engineers, wrote a blog post about using WebRTC to stream a remote server screen. Here’s how it works:

1. The service starts and listens on port 9000 by default. We can change this with a flag.
2. The service exposes two endpoints: (1) POST /session starts a session and (2) GET /screens gets the available screens from the remote server.
3. Once a screen is chosen, an SDP offer is created in the browser. For simplicity, there is a workaround in place to avoid trickle ICE.
4. Once all ICE candidates have been gathered, a POST /session request is made with the selected screen and browser’s SDP.
5. The backend inspects the offer and checks to see if the browser supports h264 baseline 3.1. If everything goes well, the server responds with an SDP answer.
6. The browser should receive and show the video stream.

Check out Rafael’s blog post to learn how to replicate this and get around a known bug in Firefox. Keep an eye out for more WebRTC tips on his blog soon!