If the participants in your study will only need to add a small number of cards (e.g. 2 or 3 extra cards for a card sort with 30 cards), you might try a workaround with placeholders. In that case you could add a few extra 'blank' cards to the set with their own unique identifiers (could be simple numbers), and instruct your participants to use them if needed and define their labels in the comments section (I know websort allows participants to leave comments during the sort, not sure about optimalsort).
So when your participants has finished sorting the prelabelled cards but thinks he misses an item, he can take a card 'BLANK CARD 1', sort it into the appropriate category, and leave a comment 'Blank card 1: item description'. Note that if you decide to do this, the standard analyses from the raw results won't be accurate (What 'Blank card 1' will be different for each participant), so you might need to do the analysis offline.
This won't work if you need to add more than just a few cards, because in that case it becomes too difficult to track what each card means.