If you check out this picture of the Brillouin zone of a hexagonal cell, you can see points like L, A and H are defined for a 3D Brillouin zone. This is because the k-path finding algorithms are generally made to work for general 3D cells, which is what SeeK-path is giving you. I'm guessing that for VASPKIT, there would be a point at which you indicate that you are simulating a monolayer, so it only gives you the high-symmetry k-points for a 2D Brillouin zone. So in short, the VASPKIT one is the one you should be using because there is no periodicity in the z-direction for a free-standing monolayer (even though, from a simulation standpoint, the DFT codes we use mostly impose periodic boundary conditions in all directions).
As a little side note that might be useful to people with similar concerns and are looking for alternatives besides VASPKIT, I like to use Pymatgen to get high-symmetry k-points, but it runs into the same problems for slab type structures. This MPInterfaces package has a nice function in this file of their GitHub repo called
remove_z_kpoints() that can be used to remove the high-symmetry k-points with a z-dimension so you're only left those that would be found in a 2D Brillouin zone.