Abstract: Spatio-temporal intelligence (STI) in agriculture relies on multi-scale and multi-temporal platforms within an integrated space–air–ground–water system, including satellites, unmanned aerial vehicles (UAVs), ground-based platforms and aquatic sensing systems. By integrating multimodal sensing, the Internet of Things (IoT), artificial intelligence and other emerging technologies, agricultural STI has been increasingly applied to regional climate monitoring, full-cycle crop management, livestock farming and aquaculture, being an important research frontier in smart agriculture. Nevertheless, effective interactions between spatio-temporal multimodal data and scenario-specific algorithms remain insufficient to support end-to-end intelligent operations. Moreover, it still lacks a systematic review of agricultural STI, its technological architecture and the representative applications. This review first examines the value chain of agricultural STI. The upstream segment comprises intelligent equipment and enabling technologies that provide the necessary hardware and software infrastructure. The midstream segment covers major stages of agricultural production, including field preparation, planting, management and harvesting, with an emphasis on information processing, data analysis, equipment operation, performance testing and maintenance. The downstream segment encompasses product storage, transportation and operational decision-making. Based on current technological characteristics and emerging trends, an overall technical architecture for agricultural STI is then proposed. Multidimensional spatio-temporal sensing and decision-making within the space–air–ground–water system are analyzed across four major production scenarios: open-field farming, orchards, protected agriculture and livestock farming. Three groups of enabling technologies—multidimensional spatio-temporal perception, efficient computing and intelligent decision-making—are examined in detail, together with their potential agricultural applications. Representative applications of agricultural STI are further reviewed, including space-based observation, space–air–ground integration, crop breeding and full-cycle growth management, livestock farming, and aquaculture. Space-based applications primarily involve satellite remote sensing for vegetation monitoring and large-scale climatic observation. Space–air–ground systems integrate information across multiple spatial layers, with low-altitude platforms linking satellite observations to ground operations and supporting soil-moisture monitoring, nutrient assessment, pest and disease detection, agricultural machinery navigation and scheduling, and microclimate monitoring and prediction. Applications in crop production encompass breeding, monitoring, management and decision-making for field and orchard crops, as well as sensor networks and intelligent control systems for protected agriculture. Livestock applications focus mainly on animal-behavior monitoring and disease prevention, whereas aquaculture applications emphasize environmental monitoring and the transportation of live aquatic products. Finally, this review identifies major challenges, including limited routine equipment-monitoring capabilities, the absence of general-purpose architectures for spatio-temporal data fusion, insufficient model generalizability, high equipment and service costs, and inadequate protection of data ownership. Future research should therefore advance toward intelligent applications covering diverse production scenarios, increasingly autonomous decision-making, and more cost-effective and efficient technological solutions. This review provides a systematic reference for developing STI-enabled smart agricultural production systems.