I would love to see a metadata source for importing from a URL, based on a "uri" or "url" identifier. It would mean, for each item metadata is being queried from take a look for the identifier. If it is present load that webpage and scan for schema.org compliant values. Then use those values to fill things in.
For example DriveThruRPG uses the
https://schema.org/CreativeWork item type for all of the things they have on the site. This would make it easy to import a title, thumbnail, rating, description, and author.
It wouldn't remove all the work of importing DriveThruRPG items, or similar things. But it would mean that if the effort were put into setting the identifier on a bunch of items everything else could be pulled in quickly and easily.