Currently, there is concern about food products to which new psychoactive substances (NPS) have been added illegally. However, rapid, high-sensitivity, and high-throughput screening methods for such harmful substances have not been reported yet. In this study, a high-throughput non-targeted screening strategy was developed based on UHPLC-Q/Orbitrap. First, optimized pretreatment methods for typical high-risk matrices were established. Second, an accurate, efficient, and scalable high resolution mass spectrometry database containing 206 NPS and related compounds was constructed. The fragmentation pathways of three different types of fentanyl-related compounds were then systematically investigated to enhance compound identification. Additionally, a non-targeted screening strategy was developed using warning ions derived from 24 fentanyl-related compounds, streamlining the identification of NPS. This strategy was applied to 100 food samples with possible illegal additives and demonstrated its effectiveness. In summary, this study provides robust technical support for regulatory authorities in combating the illegal addition of food products.