In the last decade, the focus of agent research has shifted from single agent systems to multi-agent systems (MASs). Dynamic agent coordination is one of the challenge problems of multi-agent research. One coordination problem is how to achieve agent scheduling under open dynamic environments. Petri Nets (PNs) and Coloured Petri Nets (CPNs) are system study tools that provide an appropriate mathematical formalism for the description, construction and analysis of distributed and concurrent systems. In this paper, we present a CPN based strategy to schedule and allocate new tasks to suitable agent(s) or agent combinations. In this strategy, through using CPNs to represent the dynamic statuses of agents, agent coordinators are able to check concurrent agent statuses and make correct and optimal decisions.