Abstract: The genus Fusarium is a prevalent contaminant in medicinal herbs, posing a dual threat to both the effective components and safety of traditional Chinese medicine (TCM), as well as a common pathogen in agriculture and clinical settings. Accurate identification of closely related species within this genus has always been a key challenge. In this study, we employed analysis of whole-GEnome (AGE) to conduct bioinformatics analysis on 304 genomes from 173 species of Fusarium, successfully constructing a species-specific target sequence library. The number of specific target sequences predominantly ranges from 10 001 to 50 000. To validate the reliability of the species-specific target sequences, we selected seven representative species of high concern in contaminated medicinal herbs and agricultural pathogens: Fusarium oxysporum, F. solani, F. fujikuroi, F. sambucinum, F. culmorum, F. incarnatum, and F. equiseti. Species-specific target sequence was selected from each species and subjected to detection using Sanger sequencing and CRISPR-Cas12a technology. The results showed that the species-specific target sequences exhibited high specificity, solving the identification challenge of closely related species F. incarnatum and F. equiseti. Furthermore, AGE enabled the visualization of identification results, offering an advantage for field detection. This study demonstrates that AGE can accurately identify closely related species of the genus Fusarium, providing technical support for the safety monitoring and quality assurance of medicinal herbs.