Background: https://github.com/zotero/translators/commit/ecc40d9660df478229a0552d883ba814fee25c60

Is the problem that all users use the same API key, or that the API key is published on GitHub (and present in plain text in the translator files)?

Ok, I understand that there was no real alternative option. But essentially, it results in useless scraps for anything that doesn't have an ISBN.

I used the amazon-importer for my film-collection, and now it just imports every dvd as a book, lists every contributor as a director, doesn't add any of the useful meta-data, etc.
This really should improve fast, as the amazon-import is a major feature.

To be clear, the issue is just that the product advertising API requires the same sort of API key as all their other products (e.g., their cloud services), and they have systems that (quite reasonably and helpfully) scour the web for API keys to make sure that people's keys haven't leaked out—which for many other services of theirs would be very bad indeed.

We could sign the Amazon requests via our servers without proxying them, but that'd be another service that we'd have to keep running, and even though we wouldn't log them we'd technically have access to all the URLs people were saving from, which right now isn't the case for any other translators used via the main clients.

I understand. But folks could ask Amazon to make the desired metadata available via other ways than their API. E.g. by embedding it in the pages.

I don't know if we'll be able to get the international amazon pages to work as well as they used to before. Likely not. It's a lot of work and very fragile without the API. Best idea would be to use other sources - library catalogs e.g.