slipstream

Shared distillation cache for AI agents — every fetch ~73-89% fewer tokens via a shared cache.