Abstract: Cooperation among multiple agents is a key concept that differentiates multi agent systems from other related disciplines; and it is also one of the keys to drawing multiple intelligent systems together. In this paper, cooperation processes are studied by bringing planning, competition, constraints, and coordination into a cooperation framework. In the framework, a cooperation process is divided into six stages, that is, ① producing requirement and the goal, ② planning the cooperation and solving the cooperation structure, ③ looking for cooperation partners, ④ selecting a solution, ⑤ achieving the goal, and ⑥ evaluating the result. While cooperative behaviors among multi agents are being specified, logic and process algebra methods are melted skillfully into a unified formal framework.