Finding the main article on a random web page is the hard part. Here are the two places it goes wrong most.
Failure 01
News sites with stuff inside the article.
What's there
A real article, but with embedded tweets, video players, related-story cards, and a comments section all sitting inside the same container.
What goes wrong
Lexa restyles the lot. The article gets the right treatment. So do the tweets, which were never meant to be there. The comments turn into a wall of giant text. The page feels louder, not calmer.
Failure 02
Pages with more than one article tag.
What's there
Sites that use the <article> tag for everything: the main story, the sidebar pieces, every "you might also like" card. The script can't tell which one you came to read.
What goes wrong
Lexa picks the wrong one. It might style a sidebar card as if it were the main article, or split styling across two of them. Either way you don't get the result you wanted.
Finding the right article on a random page is the one part of Lexa that's never fully done. The design decisions were easy. The content script is what taught me how messy real pages are.