Heterogeneous dual-manipulator systems combine robots with non-interchangeable capabilities (e.g., a fingered grip per and a suction cup) to enable cooperative manipulation such as handovers and staged insertions (Fig. 1). Compared to homogeneous teams, planning is affected by two prop erties: (i) actions are bound to specific robots due to capa bility constraints, and (ii) cooperation induces synchroniza tion requirements, as some actions are only feasible when robots are simultaneously available. Consequently, alloca tion, precedence, and coordination constraints are embed ded in the task structure and must be represented explicitly. In this work, we focus on the modeling layer and argue that PDDLStream [1] provides a symbolic–geometric backbone for capturing these constraints while preserving geometric executability.