JWTKey: Automatic Cryptographic Vulnerability Detection in JWT Applications

Bowen Xu, Shijie Jia, Jingqiang Lin, Fangyu Zheng, Yuan Ma, Limin Liu, Xiaozhuo Gu, Li Song

Abstract: JSON Web Token (JWT) has been widely adopted to increase the security of authentication and authorization scenarios. However, how to manage the JWT key during its lifecycle is rarely mentioned in the standards of JWT, which opens the door for developers with inadequate cryptography experience to implement cryptography incorrectly. Moreover, no effort has been devoted to checking the security of cryptographic usage in JWT applications. In this paper, we design and implement JWTKey, a static analysis detector leveraging program slicing technique to automatically identify cryptographic vulnerabilities in JWT applications. We derive 15 well-targeted cryptographic rules coupled with potential JWT key threats for the first time, and customized analysis entries and slicing criteria are identified accurately based on the observation of diversified JWT implementations, thus achieving balance between precise detection and overhead. Running on 358 open source JWT applications from GitHub, JWTKey discovered that 65.92% of the JWT applications have at least one cryptographic vulnerability. The comparative experiments with CryptoGuard demonstrate the effectiveness of our design. We disclose the findings to the developers and collect their feedback. Our findings highlight the poor cryptographic implementation in the current JWT applications.