Hello. I see that the bjt has 4 terminals, base, collector emitter and substrate. If i want to make the bjt behave as the 3 terminal device we know, which terminal do you connect it to? If i want to avoid having to connect substrates to vdd or ground, do i need to isolate the bjts? If so which one and how?

See this lateral PNP. Substrate must be connected ground. The base contact is placed on the nwell. Emitter and collector are connected to diffucions.

Can the pnp be isolated?

It is. Every nwell is isolated. The base voltage must be higher than ground.

What about npn transistors?

Anyway, BJTs are rarely used in pure CMOS technologies. They are mostly present in bandgap references. Don't use them anywhere else. If you want to use them for other applications, use IHP BiCMOS PDK

I wish to avoid body effects.

Without isolation, the body effect will harm them.

Note that the original SkyWater models for the PNP devices had 4 terminals, but the device (as provided by SkyWater) has the collector as the substrate, so it is really a 3-terminal device, and the collector of the device must always be grounded.

So you mean to say that the bjt are parasitic in nature?

Well, since there are no specific mask layers for defining a BJT, then the device is made simply from the arrangement of diffusion, well, and substrate, which are the same layers that make up a parasitic bipolar. Which is why Luis said that the bipolars on sky130 should be used only in very limited cases, such as creating a bandgap reference, and that if you want a good bipolar device, you should use a process like IHP that is a true BiCMOS process.