Thanks you for the reply - UCSC's annotation is rather muddled. I already have a lot of scripts written that "commit" me to doing it the long way, so if I understand you correctly, in my example, a mutation in position X from a vcf means position X in the forward strand, so to reconstruct the CDS, I need the reverse complement. Correct?
In other words, if I'm given coordinates 10-15 for ATGCCT, position 11 corresponds to the G in the reverse complement forward strand, not to the T (nor its complementary A).