The coastal zone, situated at the land-ocean interface, represents a complex regional system in which physical, ecological, and social subsystems are highly coupled. Under the dual impacts of global change and human activities, shoreline dynamics and ecological degradation have intensified, and the system may be approaching or crossing tipping thresholds. This study addresses the mathematical representation and classification of tipping points in coastal zone systems. Based on dynamical systems theory, we developed a unified mathematical framework that incorporates land-sea interactions and multi-scale processes. Here, we establish a six-category classification system encompassing Bifurcation-driven, Noise-driven, Shock-driven, Rate-driven, Space-driven, and Information-driven tipping points, and further elucidate the spatiotemporal characteristics and triggering mechanisms of each category. Furthermore, a classification model of tipping points was constructed using large language models and literature-based big data, and 91 typical cases worldwide are analyzed, through which the heterogeneous spatial distribution of such events is revealed. The results demonstrate that coastal tipping point research possesses both representativeness and particularity. Future research should focus on key issues of “classification-identification-response”, and advance integrated efforts in data fusion, model development, and adaptive management. Based on the above studies, we believe that future research on coastal tipping points needs to effectively integrate data, models, and knowledge, and make full use of AI and other approaches to achieve a more systematic and comprehensive understanding of these tipping points.