Thanks!
Some suggestions to clean up the code a bit & make it more robust

Thank you for your quick feedback! This is my first translator and I'm not used to Github so I apologize in advance for any silly mistakes

This all looks very good now thanks (and apologies for the delay).
Two remaining issues:

1. The root that you specify -- that shouldn't be necessary. Zotero can handle relative links and they're more robust. Could you test if all of this still works entirely without root and then remove it? (If you're used to other webscraping tools e.g. in R or python, they tend to behave less like a browser and break with relative links; Zotero doesn't)

2. Please use the minimized version of text() and attr() as e.g. here: https://github.com/zotero/translators/blob/master/Emerald%20Insight.js#L37 use all three lines -- the first one identifies, the 2nd disables linting, the 3rd is the minimized code.

I am not sure how to do without the root part because, if I understand it correctly, Zotero inserts the current url argument if I don't specify it; while it would work on the items pages or the list of publications, it would break eg on an authors' page where the current url does not match the beginning of an item's url (for example: you're on this page https://cowles.yale.edu/author/herbert-e-scarf, and select an item to import to zotero. The current url used is https://cowles.yale.edu/author and if I just append the item url for "/cfdp-2212" it throws a 404 because the actual beginning should be https://cowles.yale.edu/publications/cfdp). Same goes for the PDF, as what is scraped is an address starting at the root of the site; if I delete var root it sends a GET to the current URL (eg. https://cowles.yale.edu/publications/archives/cfm) + the scraped address whereas it should only send a GET to https://cowles.yale.edu/ + the scraped address...

Maybe I missed something because I did not take the time yet to see how Zotero connector works please tell me if you think I could do it using only relative links

Did you test? Relative links always refer to the host, i.e what you call the root and that's what Zotero uses (or certainly should use). In other words root + "/" + variable is essentially identical to "/" + variable but will behave better e.g. if URL proxies are involved or with http/https differences

Sorry I was confused it seems I got that error by testing without the initial /, in that case apparently Zotero appends the entire current URL and not just the host. It seems to work now with the initial /