Faced with increasingly complex multi-agent cooperative application, there is a type of problem that requires forming agent coalitions for different tasks, which often require resources of varying types and amounts. The traditional approach is to form disjoint coalitions for each task, each agent can join at most one coalition. In order to improve the resource utilization of agents and increase the system utility of the multi-agent system, this paper represents the problem as an overlapping coalition formation (OCF) problem and model it as a constrained optimization problem. Then an OCF method based on negotiation mechanism is developed to obtain the maximum system utility for each agent coalition, which can also satisfy the resource constraints of tasks and resolve resource conflicts of agents in different coalitions through sequentially allocating tasks. A greedy resource consumption algorithm is integrated into the OCF method. Through numerical simulations, the OCF method is validated, which can generate overlapping coalitions for different tasks without resource conflicts.