Add something at https://github.com/malxau/yori/blob/master/pkg/bld.ys1#L28 to set FDI=1 for ppc. This is because there won't be a PowerPC cabinet.dll so the only way to decompress .cab files is via statically linking the decompressor. This is available in the NT 4 SDK.
The most painful part is the debugger logic. When Yori executes a .cmd script, it runs CMD under a mini-debugger to capture the environment from the CMD process when it dies. "Real" debuggers are very CPU-specific. This one just hits behavioral quirks. As far as I can tell, x86 issues a breakpoint to the debugger and on continuation moves to the next instruction; MIPS had to be told to move to the next instruction (see https://github.com/malxau/yori/blob/master/sh/wait.c#L495 ) and I have no idea what PPC would do. It also wouldn't shock me if the memory alignment needs updating, since this code is really trying to find the environment block in CMD's memory so if any structure layout is different for PPC it might look in the wrong place.
I don't mind taking this as-is, since it's a good standalone first step. If you're looking for complete PowerPC support: