It doesn't. For some reason, which only the CoreAudio team could explain, it is necessary to use a kernel extension to capture system audio on macOS. Kernel extensions are not allowed in the Mac App Store (and rightly so), so having it just work is unfortunately not possible.
That being said, I could easily add a feature to record any other input device on the system, and then you can install the Soundflower kernel extension (or one of Rogue Amoeba's other excellent utilities) to create a loopback audio input device.
I always watch short video game replays with the audio off or on gfycat or whatever where there is no audio anyway, and for capturing bug reports or whatever I don't need audio either, so I didn't feel like I was missing much without audio for myself.
All that said, if the CoreAudio team would come to their senses and provide a userland API for doing system audio capture, I would definitely be on that.
Monosnap records video and audio, and is available on the App Store. I don't know offhand if they record off the built-in mic or if they hook the audio stream some other way, however.
To add to that, I've been on the fence about adding audio muxing. I should just do it, and folks who have a Rogue Amoeba app installed would be able to make use of it. Compared to storing the video data, the audio data is tiny, so buffering a minute of it is no problem (even if I stored it uncompressed to save CPU cycles, it would be under 10MB).
One option is to just allow people to select a single audio source, that way you don't have to do any mixing and they could use a 3rd party audio capture tool as the input source.
