This paper focusses on learning from command filtering-based adaptive neural control for non-affine nonlinear systems with full state constraints. This paper utilises a novel transformed function to convert the origin full-state constrained problem into an equivalent unconstrained one. By combining command-filtered backstepping, a novel adaptive neural control scheme is proposed to guarantee that all closed-loop signals are uniformly ultimately bounded and all states stay in the predefined bounded regions. Subsequently, by verifying the partial persistent excitation condition of the radial basis function neural network, neural weight estimates are proven to converge to the ideal weights and the convergent weights are stored in the constant values. Using the stored knowledge, a neural learning control scheme is developed for similar control tasks, which can improve control performances without the violation of full state constraints. Simulation studies are performed to illustrate the validity of the proposed scheme.