This paper proposes a trajectory prediction algorithm based on target intention recognition and distribution cropper for maneuverable targets. Different from the existing research, the trajectory prediction is decoupled into the structure of intention recognition and trajectory extrapolation to introduce intercept engagement and distribution cropper. Firstly, the intention recognition and trajectory prediction are modeled to establish the structure. The intention recognition based on intercept engagement is utilized to enhance trajectory prediction accuracy. Secondly, the distribution cropper in the graph attention mechanism is established based on the pursuit-evasion decision. Unreasonable prediction results are filtered out through the interactive distribution cropper under the intercept engagement. Thirdly, the trajectory prediction training process is supervised based on the proposed learning mechanism that incorporates intercept engagement and pursuit-evasion decision. This approach enhances neural network training performance by incorporating the adversarial dynamic between the target and the defender aircraft. Finally, the effectiveness and feasibility of the proposed algorithm are demonstrated through numerical simulations under representative intercept engagement scenarios.