Actually it's been mostly vanilla JS here, well jQuery.
We did some websites years ago, then helped make some simple casino games in JS, then made our first own big game, and didn't move to an engine because it seemed like the fastest transition.
We're also mostly in incremental genre where vanilla has it's advantages as it can make optimization far easier (and there is not that much of a focus on graphics generally), when it comes to running a ton of different numbers and UI updates, and we've sort of just evolved our own code over a few years.
If we weren't focusing on a specific niche, and having a specific marketing strategy, where how we deploy/share our games, and games having a good chunk of similar functionalities (similar UI, settings, display/UI update,...), I definitely wouldn't go vanilla again. Right now we're still aiming to focus on a specific genre where we can optimize our framework & process that gets better & faster for each game, but this is what I'm sort of putting to the test with this jam - see if we should just move to using an engine, or if our framework really has that many advantages for us.
That said, I am going to experiment with Three.js in the future, but unfortunately not too familiar with any frameworks other than reading/listening about them - i.e. pixie.js, p5.js,..
If I were focusing on mainly 2D games, I'd have definitely chosen Phaser or Godot for web - especially if I wasn't focusing on a specific niche, in which case it's best to just get to know an engine well. Hope that helps, but unfortunately can't give you much more info on other frameworks, because well - I haven't used them much. <- I don't recommend following the same path lol, unless you have a good reason. If I was just starting, I'd just try out a few different ones, see what fits me, and then focus on using one.