CVE-2026-134459 in jwt-framework
Summary
by MITRE • 11/21/2026
web-token jwt-framework and jwt-library before 3.4.10, 4.0.x before 4.0.7, and 4.1.x before 4.1.7 contain a padding oracle in RSACrypt::decryptWithRSA15() because RSA1_5 decryption lacks implicit rejection, so invalid padding, wrong CEK length, and valid padding take measurably different time. Remote attackers can submit adaptively crafted JWE encrypted_key values and measure these timing differences to recover the wrapped content encryption key.
If you want to get best quality of vulnerability data, you may have to visit VulDB.
Analysis
by VulDB Data Team • 11/21/2026
The vulnerability identified in web-token jwt-framework and associated libraries prior to versions 3.4.10, 4.0.7, and 4.1.7 represents a critical cryptographic implementation flaw within the RSACrypt::decryptWithRSA15() function. This component is responsible for handling RSAES-PKCS1-v1_5 decryption operations used in JSON Web Encryption (JWE) contexts. The core technical deficiency lies in the absence of implicit rejection mechanisms during the padding validation phase. In a secure implementation, any deviation from the expected PKCS#1 v1.5 format should result in an immediate and uniform failure response that does not leak information about which specific part of the input was invalid. However, the affected code exhibits distinct timing behaviors depending on whether the padding is completely malformed, results in an incorrect Content Encryption Key length, or passes initial structural checks but fails later validation steps. This side-channel leakage allows for a sophisticated timing attack where an adversary can exploit these measurable differences to deduce information about the plaintext without direct access to it.
From an operational perspective, this vulnerability enables remote attackers to perform adaptive chosen-ciphertext attacks against encrypted communications secured by the affected libraries. By submitting adaptively crafted JWE encrypted_key values and precisely measuring the server's response times, an attacker can iteratively narrow down the possible values of the wrapped Content Encryption Key. Once the CEK is recovered through this timing oracle, the confidentiality of all data protected under that key is compromised. This undermines the fundamental security guarantees provided by RSA encryption in JWE payloads, potentially exposing sensitive user data, authentication tokens, or proprietary information transmitted via these encrypted channels. The attack does not require high privilege levels and can be executed remotely over a network connection, making it particularly dangerous for web services relying on this library for secure message exchange.
This flaw is classified under CWE-358, which denotes an Improperly Implemented Security Check for Standard, as the implementation fails to correctly validate security parameters during decryption. Furthermore, it aligns with MITRE ATT&CK technique T1564.002, specifically Hidden File and Directory Execution via Timing Side Channels, although in this context, the side channel is used for key recovery rather than execution hiding. The vulnerability highlights the critical importance of constant-time comparison algorithms and blind RSA decryption techniques when implementing cryptographic primitives to prevent information leakage through timing variations.
To mitigate this risk, organizations must immediately upgrade their dependencies to web-token jwt-framework version 3.4.10 or later, jwt-library version 4.0.7 or later, or version 4.1.7 and above. These updated versions incorporate the necessary cryptographic hardening measures, including implicit rejection logic that ensures uniform processing time regardless of padding validity. Until an upgrade is feasible, temporary mitigations may include enforcing strict input validation at the application layer to reject malformed JWE structures before they reach the decryption routine, although this is not a substitute for fixing the underlying library flaw. Additionally, monitoring for unusual patterns in request latency could help detect ongoing exploitation attempts, though such detection mechanisms are secondary to patching the vulnerable codebase.