I'm sorry to hear about these problems, Aleksandr. Google Scholar should be quite accurate, it would be really useful to know exactly which pages are having this problem? Google Scholar is one of our most used sites, so it would be good to know of any problems with it.

I had a look at the rest that you mentioned. eLibrary looks to work well with the extension, although if you have pages where it doesn't work it would be good to get them. JSTOR and Academia seem to have similar problems: our extension looks for identifiers on the page (DOIs, ISSNs and the like) and uses that to get metadata and PDFs. JSTOR and Academia do not appear to have them. However, we do have entity identifiers which can clearly be improved for these pages, I'll report them to our developers.

Thanks for reaching out!