Abstract: Climate change is profoundly altering extreme rainfall patterns, hydrological processes in mountainous regions, and the risk structure of geological hazard risks. Rainfall-triggered geological hazards, such as landslides, collapses, and debris flows, are exhibiting new characteristics, including enhanced clustering, increasingly complex cascading processes, significant risk spillover, and greater challenges in prevention and control. Traditional geological hazard prevention and control models, which mainly rely on hazard-site investigation, empirical threshold-based early warning, and post-disaster engineering mitigation, are no longer sufficient to address compound and cascading risks under climate change, such as “rainstorm–flood–flash flood–landslide–debris flow–traffic disruption–emergency response obstruction.” This paper aims to systematically review research progress on rainfall-triggered geological hazards under climate change, and to summarize their risk evolution characteristics, mechanistic understanding, modeling approaches, and prevention pathways. Focusing on extreme-rainfall-induced geological hazards, this review analyzes relevant studies from the perspectives of climate change and extreme rainfall, risk evolution of rainfall-triggered geological hazards, event-based landslide databases and formation mechanisms, hazard assessment and intelligent prediction, and disaster-chain-based monitoring, early warning, and risk prevention. Particular attention is given to research progress in China on clustered landslides induced by heavy rainfall, debris-flow risk prediction, extreme-rainfall-induced disaster chains, and emergency risk prevention and control. The review shows that extreme rainfall promotes the evolution of rainfall-triggered geological hazards from single-site failures to regional clustering, from single-hazard impacts to cascading disaster-chain amplification, and from natural hazard processes to systemic risks by altering rainfall inputs, slope hydrological responses, deterioration of rock and soil strength, initiation of channel sediment sources, and exposure of elements at risk. Event-based landslide databases, multi-source remote sensing, machine learning and explainable artificial intelligence, physically constrained models, closed-loop early warning before, during, and after rainfall events, and disaster-chain risk identification provide important support for dynamic risk identification and prevention. Future prevention and control of rainfall-triggered geological hazards should shift from “static hazard-site management” to “dynamic risk-process prevention and control,” from “single-hazard assessment” to “disaster-chain risk prevention and control,” from “empirical early warning” to “integrated data–mechanism–intelligence-based early warning,” and from “post-disaster response” to “climate-adaptive resilience governance.” These transitions represent both an important direction for the development of geological hazard research and a practical requirement for improving the national capacity for integrated natural disaster prevention and reduction.