Liposcelis bostrychophila Badonnel (Psocoptera: Liposcelididae) is a booklouse pest that is a threat to commodity storage security worldwide. Accurate and sensitive methods of L. bostrychophila on-site identification are essential prerequisites for its effective management. Evidence suggests that L. bostrychophila contains 3 intraspecific biotypes that are morphologically indistinguishable but can be discriminated at the level of mitochondrial genome organization and sequences. The traditional molecular identification methods, such as DNA barcoding and PCR-RFLP, are instrumentally demanding and time-consuming, limiting the application of the identification in the field. Therefore, this study developed a new CRISPR/Cas12a-based visual nucleic acid system based on the mitochondrial gene coding for NADH dehydrogenase subunit 2 (nad2), combined with recombinase polymerase amplification (RPA) to accurately identify L. bostrychophila from 4 other common stored-product booklice, and also differentiate 3 biotypes of this species at the same time. The entire identification process could be completed at 37 °C within 20 min with high sensitivity. The system could stably detect at least 1 ng/µl of DNA template. The green fluorescence signal produced by the trans-cleaving of the single-stranded DNA reporter could be observed by the naked eye under blue light. Additionally, the suggested system combined with the crude DNA extraction method to extract DNA rapidly, enabled identification of all developmental stages of L. bostrychophila. With crude DNA, this novel diagnostic system successfully identified an unknown booklouse by holding the reaction tubes in the hand, thus can be considered as an accurate, rapid, highly sensitive, and instrument-flexible method for on-site visual identification of L. bostrychophila.