In this paper, we study the multi-agent distributed time-optimal formation shaping problem by proposing the time-optimal formation shaping distributed algorithm based on distributed convex alternating projection method, in which each individual only has limited communication range and sensing ability, where centralized algorithms failed. In the proposed algorithm, we model the problem as a equality-constraint distributed minimax convex optimization, which is further solved by proposing a novel virtual-equality-constraint based alternating projection method. According to the optimization results, multi-agent can achieve optimal shaping with the RVO avoidance strategy. Finally, the simulation of 100 multi-agent optimal formation is demonstrated to verify the efficiency of the proposed algorithm.