To convert byte values to code points, you need to apply an encoding algorithm, and that's kind of messy.
(I believe the rest of GNU APL kind of assumes that UTF-8 is the standard encoding used, which does make things simpler).
I have a suggestion: Make ⎕UCS support a dyadic form where the left-hand side specifies the encoding to use. I.e:
'UTF-8' ⎕UCS 99 100 101 102
Handling multiple encodings is easily done through the libiconv library. I worked on it when I made some improvements to its Common Lisp integration. It's quite simple to use.
Regards,
Elias