Hard-Coded Credentials in SciPhi-AI R2R JWT Handler

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.

Casky was already ahead

This CVE exploits attack patterns that Casky's 233matched skills already investigate — long before this vulnerability was disclosed. Claude's reasoning model maps these techniques to MITRE ATT&CK, so practitioners who ran these skills have already seen the threat behaviour in their findings.

Analysis

SciPhi-AI's R2R platform versions up to 3.6.6 contain hard-coded cryptographic secrets in the JWT Secret Handler component, specifically within the DEFAULT_BCRYPT_SECRET_KEY and DEFAULT_NACL_SECRET_KEY parameters. This vulnerability (CVSS 7.3) allows attackers to bypass authentication mechanisms by leveraging predictable or exposed secret keys, potentially gaining unauthorized access to sensitive APIs and data. Organizations deploying R2R for AI-powered applications are at immediate risk, particularly those exposing these services to untrusted networks. The vulnerability's public disclosure combined with vendor non-responsiveness creates a critical window where threat actors can systematically compromise affected deployments.

Casky's 233 mapped security skills leverage Claude's extended reasoning to detect Credential Access techniques (MITRE TA0006) associated with this vulnerability. Practitioners using the platform will identify attack patterns including: suspicious authentication token generation with non-standard entropy, JWT validation failures suggesting credential reuse, and anomalous API access patterns using weak or predictable secrets. Casky's analysis would surface indicators such as repeated authentication attempts succeeding with identical credentials, unexpected service account activity tied to default key materials, and failed cryptographic validations—all hallmarks of hard-coded secret exploitation. Security teams gain visibility into post-compromise behavior where attackers leverage stolen JWT tokens for lateral movement and data exfiltration.

Live Threat IntelligenceComing soon

Casky Risk Score

—

EPSS Probability

—

Shodan Exposure

—

GreyNoise Signal

—

Composite risk scoring from EPSS, CISA KEV, Shodan, and GreyNoise — 21 security APIs correlated into a single Casky Risk Score. Coming in Casky Pro. Join early access →

233 Casky skills cover this attack pattern

These skills use Claude AI's reasoning model to surface findings in the same attack categories as CVE-2026-105147.

abusing-dpapi-for-credential-access

red teaming · high

Run

abusing-shadow-credentials-for-privesc

red teaming · high

Run

Access with Stolen Session Cookie

cryptography · low

Run

MITRE ATT&CK Techniques

TA0006

Investigate the attack patterns behind CVE-2026-105147

Casky has 233 skills that investigate the attack patterns behind CVE-2026-105147. Run one and get CVSS-scored findings in 3 minutes.

Run the skill that detects this →

X Instagram LinkedIn