Okay. So it's a protection against browser exploits, not overreaching web APIs.

Good security has layers. That way if one falls through, hopefully the next layer will catch it.

It's more a boundary for the browser not to pass because it has no business. I've seen an alternative approach which used a separate user account for Firefox and then SSH forwarding of X.

WebTorrent is something I'm afraid to try due to the laws in the place I live and nobody answered me when I asked if WebRTC p2p connections first show a permission popup like Microphone or Speaker access. I don't know how the file access APIs work in JavaScript, but it's scary to think a random website could have a random JS snippet that uploads a file from $HOME.

There will of course be a permission pompt.

Come to think of it, why doesn't a WebRTC connection on something like appear.in prompt for Network permission?

Good to know, I might try it the next time I'm in another country.