Log inGet started
Package

turnAiKit.testSuite

Newest version

README.md

turnAiKit tests

Checks the Next Move (turnAiKit) planner against the package's sim model and a hand-written mock adapter: the same seed gives the same action and the same whole game; every personality at every difficulty ends its turn; a guaranteed kill is taken over a bigger hit on a healthier leader (normal, hard); a defensive unit moves from open ground into cover; an objective unit heads for the zone; a skirmisher backs away from an adjacent blade; a bodyguard stays by its medic while an aggressive control charges; hard beats easy in more than 10 of 20 seeded games.

Behaviour metrics: Sim.profile measures every personality x difficulty over 6 seeded skirmishes x 6 rounds (distance to enemies, share of turns ending in cover, attacks per unit-turn, distance to the objective, distance to the protected ally, distance off weapon range, share of flank shots). The suite requires each personality's signature metric to stand out from balanced at easy, normal and hard, every pair of personalities to differ clearly on some metric at every difficulty, hard aggressive to win at least 15 of 30 games against easy balanced, and every personality to win at least as often on hard as on easy against a balanced/normal opponent. Long loops yield with task.wait so a run never freezes the engine.

It does not test the demo scene or any engine rendering: the planner is pure Luau. Run it with asset.resolve("turnAiKit.tests.turnAiKit", "testSuite"):run().