This paper introduces novel schemes of continuous deterministic extremum seeking controllers based on multi-population games, designed for the solution of multi-constrained optimization problems on dynamical systems with a multi-agent system (MAS) structure. In this way, we consider different cluster of agents, interacting by means of different cost functions, which in general depend on the states of all the other agents. The agents of the same cluster aim to simultaneously maximize their common cost function, whose mathematical form is unknown, and which is only accessible by measurements. The optimization is carried out under different types of constraints: coupled or decoupled among clusters, describing multi-resource allocation problems and market share competition problems. The implementation of the algorithms is illustrated via simulations.