Abstract: Driven by increasingly frequent extreme weather events, rainfall-induced clustered landslides have become more common in the mountainous regions. It poses a serious threat to the safety of people's lives and property and regional economic and social development. In July 2025, the Beijing–Tianjin–Hebei region experienced a prolonged extreme rainfall event, which triggered numerous landslides in the mountainous area along the Beijing–Chengde border. In this study, a total of 7,907 rainfall-induced landslides interpreted from this event were used as the research dataset, to explore the driving mechanisms of various factors in the assessment of rainfall landslide risk. Eleven influencing factors were selected, including elevation, slope, aspect, plane curvature, terrain relief, land cover, stratum, distance to faults, distance to roads, cumulative rainfall, and maximum daily rainfall, construct the XGBoost rainfall landslide risk assessment model, and conduct interpretability analysis combined with the SHAP method. The results show that the model achieved AUC values of 0.942 and 0.932 for the training and validation datasets, with good recognition accuracy and generalization ability. SHAP analysis revealed that maximum daily rainfall and cumulative rainfall were the most critical triggering factors for landslide occurrence during this event, both exhibiting pronounced threshold effects. Factor interaction analysis revealed the coupling and synergistic effects among various environmental factors, and the coupling of rainfall factors with topography and stratigraphic conditions further amplifies landslide risks. There is a coupling mechanism between the key controlling factors of landslide occurrence under extreme rainfall conditions. The XGBoost-SHAP model can effectively identify landslide high-risk areas and explain the role of factors in the model. The research results offering insights for risk assessment and disaster mitigation in the mountainous regions of Beijing–Tianjin–Hebei.