Can you built BJT on this process btw ? (I mean purposefully, not parasitic ones :p)

I don't think so. PDK doesn't indicate BiCMOS nor does their process info show BiCMOS for 130 nm. https://www.skywatertechnology.com/technology/mixed-signal-asic/

The not-yet-ready primitive device library has several NPN and PNP devices. Because the process has a deep n-well, it is possible to get a decent NPN with ndiffusion over pwell in deep n-well. The PNP has the collector tied to substrate.

Very interesting, I haven't seen a foundry model an npn using deep n-well. And I did not know there was a deep n-well in this process; good to know. Will be curious to see the performance when it's ready. When you say "several" devices, do you mean various fixed pcell sizes? Or are there different types, such as lpnp also? I saw on a spreadsheet linked in general that these are only different variations and not different types. So either VPNP or VNPN.