I tried doing it by getting the vector from the position of the corner to the position of the face, but those would point to the middle of the face and it broke down when the face was really long. And they weren’t really splitting the corner into two perfectly equal parts anyway.
EDIT :
Just realised you want to split the corner in equal parts, here it’s just averaging the direction of each edges and by giving a second look I’m not even sure it works
Wow, so many solutions! They all work, except in an edge case when the face is completely mangled and has a corner pointing inward. Doesn’t bother me, if you have geometry like that, you have bigger problems.
Expanding on @higgsas’ method, chirality (or winding order of faces) is preserved with the Offset Corner in Face node so a cross of the 2 edge vectors will be in the opposite direction with the face normal if the angle is a reflex angle (i.e. the dot product will be negative).
I’m pretty sure I didn’t use that word correctly… It may only apply to chemistry but for some reason I have associated it with “handedness” (incorrectly I think now). Maybe forget it and just use “winding order” instead