What I've already tried:
using a cloud function with link-preview-js,
using a headless browser in a cloud service (puppeteer and some others),
scraping the embed page for the html title
Every time I only get the same generic metadata or nothing.
Anyway, thanks for investingating. If it's out of scope for what you'd be willing to help with, that would be quite sad news to me. I hoped the issue with Open Graph was not intended. I've been trying to emulate the twitter bot, but it is quite difficult and my expertise (nor cursor AI's) can crack it.