Since low-Earth-orbit (LEO) heterogeneous satellite constellations have the characteristics such as large scale，high orbital dynamics，and strong resource heterogeneity，traditional centralized management approaches for LEO heterogeneous satellite constellations is facing critical challenges such as dynamic topology management difficulties，inefficient multi-satellite coordination，and cross-protocol interoperability barriers.To address these challenges，the task management architecture and models for LEO heterogeneous constellations are systematically studied in this paper.A multi-agent distributed task management framework is proposed based on a “centralized + distributed execution” hybrid paradigm.This framework leverages high-precision relative navigation，autonomous control，and inter-satellite communication to enable stable formation maintenance and dynamic reconfiguration，thereby enhancing the mission flexibility (on-demand formation adaptation)，improving the system reliability (through redundancy)，and reducing the operational costs (via small-satellite networking)，while significantly increasing the collaborative efficiency.Furthermore，an event-driven dynamic cooperative scheduling algorithm is proposed，which consists of three core modules:event detection，dynamic task allocation，and real-time optimization.The proposed algorithm demonstrates superior responsiveness，adaptability，and resource utilization efficiency，making it particularly suitable for highly dynamic and uncertain operational environments，while ensuring rapid response and high execution precision.Building upon this foundation，an intelligent planning roadmap is presented for LEO heterogeneous constellations from the perspectives of user interfaces and information processing.Finally，some development recommendations are given，including formation maintenance strategies，standardization of space communication protocols，and integration of artificial intelligence technologies.