In a tabulae dataset, the directory orthography contains a file alphabet.fst where the alphabet for your corpus is explicitly enuerated.

This is the one component of a tabulae dataset that is not recorded as a delimited-text table. Instead, alphabet.fst is directly written in the notation of the Stuttgart FST toolkit.

Several model alphabets are included in the template that tabulae installs with the sbt corpus task. You can choose one of these and copy it to alphabet.fst.