MECHANICS
control-lock sim policy reads threat surface data (WS8.4)
What changed
rankCard for the control-lock sim policy now reads state.threatPhases: when the upcoming telegraphed run carries a rider, it ranks the BACKFIRE punish highest (no roster candidate erases a rider outright); otherwise it ranks STAGGER rung-denial highest, with lock_stance as a certainty tiebreak.
Why it changed
CRITIQUE finding [MED] "control-lock sim policy is threat-blind" — surface variety already existed in the data (WS8.2's STAGGER/BACKFIRE/lock_stance split) but the decision layer picked the same control card against every threat shape regardless.
+function controlSurfaceBonus(s: CombatEncounterState, card: CombatCard): number {
+ const source = getCardById(card.id);
+ if (!source) return 0;
+ const staggerRungs = source.specialMechanics
+ ?.find((m): m is { kind: 'stagger'; rungs: number } => m.kind === 'stagger')?.rungs ?? 0;
+ const locksStance = source.specialMechanics?.some(m => m.kind === 'lock_stance') ?? false;
+ const backfire = source.combatEffects?.find(e => e.effectId === 'debuff_backfire');
+ const backfireValue = backfire ? (backfire.intensity ?? 1) * (backfire.duration ?? 1) : 0;
+
+ if (upcomingThreatHasRider(s)) return backfireValue * 100;
+ return staggerRungs * 1000 + (locksStance ? 10 : 0) + backfireValue;
+}
@@ rankCard: (s, card) => {
if (isControlClass(card)) {
- return (isNewStatus(s, card) ? BAND_PRIMARY : BAND_SECONDARY) + card.bottomDamagePreview;
+ return (isNewStatus(s, card) ? BAND_PRIMARY : BAND_SECONDARY)
+ + card.bottomDamagePreview + controlSurfaceBonus(s, card);
}
commits d0d83e06 · e · 159 · ced · b · ad · this · de · — · ee · e · fa · cad · be