the constraint on no transforms makes this way harder than it looks. i've always relied on
transform: rotateX(60deg) rotateZ(45deg)
to handle the perspective, so relying solely on clip-path for the depth of the walls feels like a nightmare for managing the grid alignment. it's gonna be tricky to keep the edges seamless w/o that 3d rotation logic. i wonder if you'll need to use pseudo-elements heavily to simulate the side faces of the walls? how are you planning to handle the shadows without any depth manipulation? ❓