Just wondering, are you planning for integration with gnugo or similar, or do the problems cover all possible (wrong/bad) moves (ok all possible most likely not possible, but if there are comments why other fail would be nice)? In many chess problems at least why something is not the best move is not always obvious (even though much easier to test all moves), especially in problems that are not of the life/death variety, but advantageous tactical position etc, so playing with an engine to see other outcomes is helpful.