The new Extractions stuff for uploading of files and datasets and extraction of metadata using Versusu also contain implementations of features that would be cool to have for regular curation and extraction as well (e.g. upload of a file straight from a URL).