Assembly language! It has a custom 8-bit fantasy architecture called the F8B25 (yeah, not the Fairchild F8!). It has sixteen 8-bit general-purpose data registers/accumulators, a 64K address space, and a pretty basic instruction set (which honestly shares some similarities to the 6502 and some RISC-type designs, but is more limited in the context of addressing modes).
Code density is just low to be honest. They sure had enough spare opcode space for add/subtract without carry, and a way to get the stack pointer value (which is certainly a must-need thing other 8-bit ISA's have, otherwise stack-based arguments is difficult, and multi-stacks are impossible).
Also, note that the manual has a few typos here and there. I downloaded the source code, and confirmed that some opcodes are misaligned, "PSH I" exists and ALS appears to be broken. I think "CMP R, I" might be broken too, in that it doesn't set the CARRY flag if the immediate operand is 0 (always clears it, since the AND-mask and +1 add are backwards), which may indicate the code that there was a borrow even though any unsigned value minus 0 doesn't produce one.
Also, I don't think the OVERFLOW flag's detection handles all cases correctly, since it literally just detects if the MSB of the destination register is flipped by an ADD/SUB/CMP.