Abstract: Planar four-bar linkage is a common transmission mechanism. The optimal design of mechanical parameters is an important way to achieve the best dynamic performance. In this study, a multi-agent chaos bird swarm algorithm (MACBSA) is proposed to deal with the mechanical parameters' optimization problem. The proposed algorithm combines the action policy of agents in a multi-agent system and a chaos search strategy with an evolutionary process in a bird swarm algorithm (BSA). The competition and cooperation scheme between agents improves the information interaction and learning ability of individuals. It also enhances the diversity and information feedback within the swarm. In addition, the chaos search helps the algorithm break away from the local optimum. Lastly, the proposed algorithm is tested on four benchmark functions and then applied to optimizing the design of the planar linkage mechanism. Simulation results reveal that the algorithm has more advantages over seven other algorithms in precision, convergence speed, and robustness. It is more suitable for solving such mechanism optimization problems.