You'll need to get the metadata either via the ANDS ORCA system or from an OAI-PMH feed. The first option is easiest: (thanks to Grant from Flinders for providing the details below)
The downloaded file will contain several collection records so you'll need to break it up into 1 collection per file. The python script below should do this for you (but you'll want to check it first). You'll also notice that the file will break out the different RIF-CS object types - this is handy if you have a mix in the XML file. Don't forget that you only want to import COLLECTIONS into ReDBox.
A utility script for splitting up the file is provided at https://github.com/redbox-mint-contrib/config-samples/blob/master/util/rifsplit.py:
Once you've run the script across your XML export you should have a series of files in the "output/" directory. Each filename is prefixed with the object type and you only need those with "collection_".
Documentation > System Administration > Administering ReDBox > Loading data into ReDBox via Alerts >