Not just CSS, but every resources linked from the HTML document, which includes images and potentially audio and video, if you want to archive them.

Yeah i forgot, i also have to edit the CSS to change the image urls for example right?

AFAIK, most archiving service strip the Javascript. that is what i was going for, sounds like the easiest way, The issue i identified is that i probably have to do this in a whitelist fashion, since javascript can hide in a lot of places ( or it it just a few?)

I actually searched but most answers where about using other tools or already made tools to do so.

But i'll take a look at it again.

Deviations of the norm put you on a list. Good point, and considering this actually makes the solution easier.

How viable would it be to collect other's habits ( say view most popular videos etc...) and generate searches for those and related content ?

Should be easier than coming up with a random "normal" search pattern, and would be more likelly to look normal because it's a copy.