Scholix does not recommend any specific protocol to expose links, although, any community standard is strongly encouraged. As to ScholeXplorer, its harvesting layer bulk-collects links from external data sources via public APIs. OAI-PMH is preferable, but also REST APIs can be accepted. The option of “incremental” harvesting is strongly encouraged (not mandatory), for example allowing to search links by "last date of indexing": http://www.mydomain.eu/scholix?lastIndexingDate=yyyy-mm-dd). If Scholix links are returned with all metadata fields, including the optional ones, the APIs above are enough. If instead, the links are limited to the mandatory fields, then a "resolution" API would be required: given a PID of an object the API returns its full metadata record.