MAMO introduces a multi-agent reinforcement learning approach to address the challenge of balancing cost minimization and constraint satisfaction in dynamic environments. It decouples task execution from reward weight selection, treating the choice of weights as a learning problem to enable more autonomous and robust solutions.