Tox isn't viable
The tox community has a bad history, and its current development culture has problems too. I don't want to drudge all of this up here. It's not hidden. You can find old github issues if you search for them.
There are also technical issues - it is unaudited, the spec is incomplete and out-of-sync with development, and there is only one real implementation.
Further, on a protocol level, it makes no effort to hide IP addresses from other users so despite its onion routing and e2ee, your are broadcasting your IP(s) to the network, and the maintainers have made no effort to fix or otherwise account for this beyond the dismissal to "use Tor".
The original developers/designers are long gone, and as far as I can tell, none of the current active developers have the domain-specific knowledge to make large protocol improvements.
As it stands, even the addition of basic protocol features that you would want in a functioning and maintainable project, which are missing, such as protocol versioning, are stalled by disfunctional social dynamics.
I was a heavy tox user for the past 2 years, and attended multiple developer calls. I've programmed tox bots, contributed to qTox, and first used tox about 8 years ago.
I was finally able to keep tox turned off the past few months since my last contact on there moved to another platform.
I will end this with the one thing I truely like about tox, or rather libtoxcore, which is that the library has almost no dependencies, and it's overall pretty comprehensible C code. That's quite a nice accomplishment and largely what I'd want from a good messaging library.