In this paper, a state-constrained safety adaptive trajectory tracking control scheme for a hovercraft with enhanced prescribed performance is studied. Firstly, considering the position error, a novel pair of prescribed monotone tube boundary functions combined with a sliding mode is introduced to obtain virtual velocity control laws, and the position error is simplified into an equivalent unconstrained control system by adopting the transformed error function. Unlike traditional prescribed performance control, the new function has a quantitative relationship between (transient and steady-state) control performance and some practical user-defined metrics such as overshoot, precision, and convergence time, and is less conservative. Therefore, this method is convenient for design and engineering practice. Secondly, in order to solve problems of velocity safety constraints, a novel asymmetric integral barrier Lyapunov function (AIBLF) has been adopted to limit velocities of hovercraft within the asymmetric safety constraints. Moreover, a bioinspired neurodynamic model is introduced to handle differential explosion of virtual control laws. For the sake of estimating the unknown terms such as the control system uncertainties, the neural networks (NNs) are utilized. Total control system is ultimately uniformly bounded according to Lyapunov stability theories. Effectiveness and superiority of the proposed control scheme are verified by comparative simulation.