Move all your pieces to the opposite triangle
This is a two-player version of Chinese Checkers on the standard star layout with the four unused points left off: a 61-hole hexagon with a 10-hole triangle at the top and another at the bottom, each joined to the hexagon along its four-hole edge. Each side has ten pieces. Your light pieces start in the top triangle and must all reach the bottom one; the AI's dark pieces travel the other way.
Click one of your pieces to see every hole it can reach, then click where it should go. A piece can step to any of the six neighbouring holes, or make a chain of jumps, each one over a single adjacent piece of either colour into the empty hole directly beyond. Nothing is ever captured. The first player to fill the opposite triangle wins; under a common house rule, an opponent's piece still sitting in its starting triangle counts as filled, so nobody can block a win by staying home.
A two-player race on the standard star board with the four unused points left off: a 61-hole hexagon with one 10-hole triangle at the top and one at the bottom, each attached along its four-hole edge just as on the boxed game. You move first, sending your ten light pieces from the top triangle to the bottom one, while the AI's dark pieces head the other way. Selecting a piece highlights every hole it can reach with a single step or a chain of jumps over adjacent pieces of either colour. Nothing is captured, and the first side to fill its target triangle wins.
The AI always takes a winning move when it has one. Otherwise it scores positions as a race distance: rows still to travel to the far tip of its target, the distance from each outside piece to the nearest empty goal hole, and an extra charge for its rearmost piece so none is left stranded. The deepest holes count as the shortest distance, so it fills its target from the tip forward and finishes cleanly. Long jumps cut the distance most, so whenever you build a line of pieces through the centre, expect it to use that line too.
Easy adds randomness to every move's score and sometimes plays any forward move, so it wastes turns and misses long chains. Medium always plays the move that cuts its race distance the most, one move at a time. Hard looks ahead: for its most promising moves it predicts your strongest reply and its own best follow-up, so it avoids opening ladders for you and will block a lane when that costs it little. In testing, Hard beat Medium in the large majority of games, and Medium beat Easy just as clearly. AI Training removes the guesswork about your reply: for each of its 24 most promising moves it checks your five strongest answers and assumes you will choose the one that hurts it most, then picks the move that holds up best, with no randomness and in about a second at most. It beat Hard 18-9 in testing, with five more games still unfinished after 500 moves.
For the complete rules, the deeper strategy and the history of the game, read the full Chinese Checkers guide, or browse guides for all 33 games.