This paper constructs a comprehensive low-carbon policy intensity index for China's manufacturing industries from 2007 to 2022, addressing the limitations of existing studies that often use proxy variables or composite indices at the national level. The authors develop a phrase-oriented NLP algorithm and text-based prompt learning to quantify policy intensity, which is measured by policy level, objective, and instrument. The dataset includes 7282 policies at the national, provincial, and prefecture levels, and it is organized in two formats (.dta and .xlsx) for multidisciplinary researchers. The policy intensity is aggregated at different levels (national, provincial, prefecture) and categorized by four objectives (carbon reduction, energy conservation, capacity utilization, technology) and three instruments (command-and-control, market-based, composite). The study aims to deepen the understanding of low-carbon policy intensity and provide a valuable resource for future research on low-carbon policies and their impacts. The methodology combines manual labeling and prompt learning to predict policy intensities, ensuring high accuracy and reducing human bias. The results show strong cyclical trends in national-level policy intensity and highlight the importance of command-and-control policies before 2015, followed by market-based and composite instruments after 2015. The dataset and methods contribute to a more nuanced understanding of low-carbon policy effects and can be used to merge with macro and micro data for extended analysis.This paper constructs a comprehensive low-carbon policy intensity index for China's manufacturing industries from 2007 to 2022, addressing the limitations of existing studies that often use proxy variables or composite indices at the national level. The authors develop a phrase-oriented NLP algorithm and text-based prompt learning to quantify policy intensity, which is measured by policy level, objective, and instrument. The dataset includes 7282 policies at the national, provincial, and prefecture levels, and it is organized in two formats (.dta and .xlsx) for multidisciplinary researchers. The policy intensity is aggregated at different levels (national, provincial, prefecture) and categorized by four objectives (carbon reduction, energy conservation, capacity utilization, technology) and three instruments (command-and-control, market-based, composite). The study aims to deepen the understanding of low-carbon policy intensity and provide a valuable resource for future research on low-carbon policies and their impacts. The methodology combines manual labeling and prompt learning to predict policy intensities, ensuring high accuracy and reducing human bias. The results show strong cyclical trends in national-level policy intensity and highlight the importance of command-and-control policies before 2015, followed by market-based and composite instruments after 2015. The dataset and methods contribute to a more nuanced understanding of low-carbon policy effects and can be used to merge with macro and micro data for extended analysis.