The AGA mod was the first idea, just a simple way to get most repetitions right. It is not perfect (bent4 with double ko seki, some remaining superko flaws, etc), but still seem a huge improvement.
The current approach with minimal superko is not necessarily “fully formed” or final either. It is hard to get even just all three of “1 eye flaw”, “double ko seki” and “0-sided ko” correct (and moonshine oc). Minimal superko seems to work, but this may be partly coincidence. Actually (in light of 0-sided kos), even the human logic is not completely clear about what would make a final position with normal ko! (and hopelessly incorrect with regular superko) Such theory would ideally need research from several people.
Write down basic ko with threefold repetition - as if would it fit into 1-2 lines? Maybe something like:
No capture of a single stone that just captured a single stone. If a board play creates the same position for the third time, the game ends with a win for the side who captured more stones since the last occurrence, or a draw if equal.