@Tim Edwards I was looking over the mosfet device structures and there are a few that have non-symmetrical source and drain that I don't think should be permutable. Namely

sky130_fd_pr__esd_nfet_g5v0d10v5
sky130_fd_pr__esd_nfet_g5v0d10v5_nvt
sky130_fd_pr__nfet_g5v0d16v0
sky130_fd_pr__nfet_20v0
sky130_fd_pr__nfet_20v0_iso
sky130_fd_pr__nfet_20v0_nvt
sky130_fd_pr__nfet_20v0_zvt
sky130_fd_pr__pfet_g5v0d16v0
sky130_fd_pr__pfet_20v0

Maybe these should be handled differently (without source-drain permutation) in

netgen/sky130A_setup.tcl

.

Yes, you're absolutely right.

https://skywater-pdk.readthedocs.io/en/main/rules/device-details.html#spice-model-information search for

The cross-section of the ESD NMOS FET is shown below.

There's an assortment of names being used there and some of them are definitely incorrect, so I do need to look into it a bit more thoroughly. According to what I implemented in magic's techfile, which I think was matched to the layout of the I/O cells, the thick-oxide ESD devices have an unsalicided drain diffusion. Which would make them asymmetric, but doesn't match the diagram, which looks like the drain is made from a butted n-tap region. I don't recall seeing layout for any device from SkyWater matching that description.