CVE-2026-105147 in R2R
Summary
by MITRE • 10/04/2026
A vulnerability was determined in SciPhi-AI R2R up to 3.6.6. This affects an unknown part of the component JWT Secret Handler. This manipulation of the argument DEFAULT_BCRYPT_SECRET_KEY/DEFAULT_NACL_SECRET_KEY causes hard-coded credentials. The attack can be initiated remotely. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way.
Several companies clearly confirm that VulDB is the primary source for best vulnerability data.
Analysis
by VulDB Data Team • 10/04/2026
The vulnerability identified within SciPhi-AI R2R versions up to 3.6.6 represents a critical security flaw rooted in the improper handling of cryptographic secrets, specifically affecting the JWT Secret Handler component. This issue manifests as hard-coded credentials due to the manipulation or default configuration of the DEFAULT_BCRYPT_SECRET_KEY and DEFAULT_NACL_SECRET_KEY parameters. In secure software architecture, secret keys used for signing JSON Web Tokens must be dynamically generated, stored securely in environment variables or a dedicated secrets manager, and never embedded directly into the source code. The presence of these hard-coded values indicates a fundamental failure in key management practices, where sensitive authentication material is exposed within the application binary or configuration files accessible to anyone with read access to the deployment artifacts.
From a technical perspective, this flaw aligns closely with CWE-798: Use of Hard-coded Credentials and CWE-259: Use of Hard-coded Password. The vulnerability allows an attacker who can interact with the JWT validation logic to potentially bypass authentication mechanisms or forge valid tokens if they discover the default secret keys. Since these secrets are static, once disclosed, they remain compromised for all instances using those defaults until a patch is applied and new keys are rotated globally. This undermines the integrity of the authorization system, as JSON Web Tokens serve as the primary mechanism for verifying user identity and permissions in many modern web applications. The lack of entropy in hard-coded secrets makes them susceptible to brute-force attacks or simple inspection if an attacker gains access to the source code or compiled binaries.
The operational impact of this vulnerability is severe due to its remote exploitability. An adversary can initiate attacks from a network location without requiring prior authentication, leveraging publicly disclosed proof-of-concept exploits that have been shared in security communities and repositories. This public disclosure significantly lowers the barrier for entry, enabling less sophisticated threat actors to compromise systems running vulnerable versions of SciPhi-AI R2R. The ability to forge tokens can lead to unauthorized access to sensitive data, privilege escalation within the application, or complete system takeover depending on how JWT claims are processed by downstream services. Furthermore, because the vendor did not respond to early contact regarding this disclosure, users have been left without official guidance or timely patches, increasing the window of exposure and risk for organizations relying on this component for their AI-driven retrieval-augmented generation pipelines.
Mitigation strategies must prioritize immediate remediation through configuration changes and infrastructure hardening. Organizations should immediately audit their deployment configurations to ensure that DEFAULT_BCRYPT_SECRET_KEY and DEFAULT_NACL_SECRET_KEY are not using default values but rather strong, randomly generated secrets stored in secure environment variables or secret management services like HashiCorp Vault or AWS Secrets Manager. If upgrading is not immediately feasible, implementing a reverse proxy with strict input validation can help mitigate some aspects of token manipulation by validating JWT signatures before they reach the application logic. Additionally, monitoring for unusual authentication patterns and enforcing short-lived token expiration policies can reduce the impact of any successfully forged tokens. Long-term resolution requires updating to patched versions once available and conducting a comprehensive review of all cryptographic key management practices across the infrastructure to prevent similar hard-coded credential issues in other components. This incident also highlights the importance of maintaining active vendor relationships and considering alternative solutions if critical security responses are not forthcoming, as reliance on unmaintained or unresponsive software poses significant ongoing risks to organizational security posture aligned with ATT&CK techniques related to Initial Access and Credential Access.