Background: Antimicrobial resistance in Neisseria gonorrhoeae is an urgent public health threat. Resistance-guided therapy can assure appropriate treatment and reintroduce alternative therapeutic options by identifying genetic predictors of resistance. Mosaicism at codons 375-377 of the penA gene are associated with cefixime resistance. Rapid, field-deployable assays for predicting cefixime susceptibility are lacking. Methods: We used a machine-learning algorithm to develop a CRISPR Cas13a-based assay to detect the absence of mosaicism at codons 375-377 of the penA gene combined with isothermal amplification in a single reaction. We integrated the assay onto a portable fluorescence-based platform. We evaluated performance using cultured isolates and compared results with PCR genotyping and phenotypic antimicrobial susceptibility testing. We also assessed feasibility of reagent lyophilization for cold-chain-independent deployment. Results: Among 40 N. gonorrhoeae isolates, the Cas13a penA assay demonstrated 100% concordance with PCR genotyping and 92{middle dot}5% (95% CI 79.6-98.4%) concordance with phenotypic cefixime susceptibility. Median time to detection was 12 minutes (IQR 5 minutes). The lyophilized detection system detected all 12 isolates with a median time to detection45{middle dot}0 minutes (IQR 40-45) compared to 45{middle dot}0 minutes (IQR 35-50) for the positive aqueous control, although peak fluorescence was higher for the aqueous control (p<0.01). Conclusion: The Cas13a assay was rapid and demonstrated strong correlation with genotypic and phenotypic cefixime susceptibility in N. gonorrhoeae, while a lyophilized assay retained functionality.