Abstract: In recent years, extreme heavy rainfall has triggered numerous cluster landslide events in the fully weathered granite areas of southern China, causing severe losses. Factors such as regional extreme weather, thick weathered granite layers, and non-standard slope cutting practices have led to frequent occurences of cluster landslides. However, the developmental characteristics and conditions conductive to these landslides are not well-understood, posing challenges to geological disaster monitoring and prevention. This paper focuses on the granite area of Beiliu City in Guangxi as the study area. The study discuss the developmental characteristics, spatiotemporal distribution patterns, failure modes, and sensitivity of disaster-breeding conditions of rainfall-induced cluster landslides through an integrated approach combining field investigations, remote sensing interpretation, and XG-boost modeling. The results show that two rainstorm events in the area triggered 1,670 landslides, mainly of small to medium scale, accounting for 72.57% of the total occurrences. The cluster landslides primarily developed in cut slopes with elevations less than 300 meters and slope angles greater than 75°, exhibiting characteristics of wide spatial distribution, small individual scale, high frequency, immediate response to rainfall, and strong explosiveness. Three main types of failure modes were identified: shear-sliding, toppling-sliding, and flow-sliding (debris flow). The primary disaster-breeding conditions were identified as distance to residential areas (proportion contribution of 22.22%), slope (15.25%), elevation (14.47%), and normalized difference vegetation index (12.01%) in the granite areas. This study provides a basis for the research of rainfall-induced cluster landslides in granite areas and is of great significance in disaster prevention and mitigation.