In a project-driven supply chain, the project schedule and material supply influence one another. The effective decision-making process between the project manager and the suppliers can promote the flexibility and competitiveness of supply chains. However, due to their incompatible objectives, the suppliers are reluctant to disclose private information. By incorporating information asymmetry, we build a model to describe the decentralized decision-making process. The project manager does not know the lead time and the production/transportation cost of the material suppliers accurately. To build an effective alliance in the supply chain, different contracts are considered to provide a positive or negative incentive for the suppliers, including a non-financial incentive contract with continuous orders. Then, we present a framework that integrates the agent-based approach and evolutionary algorithm. In the framework, the agents not only negotiate with each other to complete a solution but also jointly evaluate the solutions generated by the evolutionary algorithm. Finally, an experiment is conducted to compare the agent-based approach and the classical NSGA-Ⅱ under information symmetry. The results show that the gap between the algorithms is acceptable, especially for a large project. The results also show that the non-financial incentive contract is beneficial to all the players in the supply chain.