Ah, okay, I just saw this post and so I see that's just not how Kria works. It's not key-based, it's interval-based.
That is, it's not designed for parallel key transpositions, but it is possible to effectively change keys by changing the mode. In other words, let's say my VCO is tuned to C so that 0 plays the low C note. I then start with the major scale (W-W-h-W-W-W-h) preset with root on the first block to play a C Major scale. Then, to transpose to A major I would change the scale to a Phrygian mode (h-W-W-W-h-W-W) and move the root to the second block to transpose the C up to C# (which is the third degree of A major). If I don't move the root, it would play Ab major (C Phrygian = Ab major).
I think it's important to recognize that with this approach, transpositions will be modally recontextualized. That is, the lowest note of a sequence, which starts off as the root note of the melody, would no longer be the root note of the new scale once transposed. In the previous example of C major to A major, the lowest note would be the third of the parent major scale, or the root of the Phrygian mode.
So in figuring out how to best leverage Kria's strengths while attempting to realize my compositional goals, I would suspect that a post-Kria/Ansible transposition (and/or quantization) could be useful. In this approach, I would regard Kria/Ansible as being modal with a fixed root and would perform key transpositions after the fact with, for example, O_C's Quantermain which has the ability to shift the root, and octave, as well as the scale.
I'm just thinking out loud here in case anyone wants to offer any new ideas or point anything out I may have missed.