<p>In this paper, a cooperative learning control scheme is presented for multi-agent systems with time-varying full state constraints. A new nonlinear mapping is proposed and the constrained tracking errors are transformed into equivalent unconstrained ones. Using the Lyapunov method, the stability of the system is verified while ensuring the time-varying full state constraints remain unviolated. This paper proves that the cooperative persistent excitation condition is satisfied under the time-varying full-state constraints. Then, the distributed cooperative learning control scheme is employed by establishing the communication topology among update laws of neural network (NN) weights. Thus, all estimated weights of NNs can converge to their optimal values with small errors. Further, the learned NNs can be used to construct the experience-based controllers to control same control tasks. Finally, we provide two examples to show the advantages of the aforementioned control schemes.</p>