is this just a total guess? the repository says it's forked from KhiCAS, which is a ti84 port of Giac/Xcas, which can do full Risch integration if necessary
https://github.com/KhiCAS/ti-ce-giac/blob/main/kintg.cc#L288...
https://github.com/KhiCAS/ti-ce-giac/blob/main/kintg.cc#L288...
https://github.com/tommyb3ll/ti-ce-giac https://github.com/tommyb3ll/ti-ce-giac/blob/main/kintg.cc#L...
why do you think it's unlikely? if KhiCAS does it, why wouldn't this fork?
Parts of it looked very much like pmint's core tactic of matching coefficients (given a form)