@User: Huh. I was finding 20V devices like that in netlists that I was trying to extract from the I/O pads yesterday, but I assumed it was due to local changes I was making in the techfile. But note that both of those netlists were extracted from magic views; the only difference is updates to the PDK since the sky130_fd_sc_hvl library was last built. So something that changed recently in the techfile is causing layer changes when reading the GDS which leads to incorrect extraction. I'll look at that right now, especially as it has an immediate effect on what I'm trying to do today with the I/O cell extraction.