To address the challenge of balancing computational efficiency and scheduling quality in task scheduling for multi-agile earth observation satellites, a two-stage intelligent constructive method based on satellite prioritization and integrated single-satellite scheduling is developed. The multi-satellite scheduling problem is decomposed into two stages: satellite scheduling sequence optimization and single-satellite task scheduling optimization. In the satellite scheduling sequence optimization stage, satellite priorities are determined by jointly considering the total benefit of observable tasks and the average length of visibility time windows, thereby enhancing the matching efficiency of critical resources. In the single-satellite task scheduling optimization stage, an ensemble insertion algorithm is designed by combining the minimum insertion cost and minimum idle time criteria. This algorithm, combined with a scheduling relaxation mechanism, enables efficient construction and local optimization of task sequences. Experimental results demonstrate that high-quality scheduling solutions are generated within seconds across scenarios with varying numbers of satellites and task scales. The proposed method outperforms advanced question-and-answer and centralized algorithms, while exhibiting scalability and real-time responsiveness. It is well-suited for imaging task scheduling in large-scale satellite constellation environments.