In a multi-player urban planning process, the outcome of any individual decision of the actors is uncertain until a state where the plan is satisfactory for all. To support the plan generation phase, this paper develops a generic multi-agent system, in which agents represent particular land-uses. In the system, agents higher in the hierarchy have priority over agents lower in the hierarchy to claim units of land. This one-direction claim process may result in a plan that is not optimal for every agent. The system, therefore, allows agents to revise their plans in an iterative procedure. A case study illustrates centralized, semi-centralized and decentralized solutions for a plan area based on the outcomes of different strategies used by facility agents (retail, green, schools) and a housing agent. The results show that the proposed system is able to generate rational and realistic plan alternatives for new residential areas.