The Omnivore API (packages/api) before the fix in commit abf53d6 contains an authentication bypass in Apple sign-in token verification. The decodeAppleToken function extracted the 'alg' field from the attacker-supplied JWT header and passed it as the sole allowed algorithm to jwt.verify(). Using jsonwebtoken v8 (which does not validate key/algorithm compatibility), an attacker can set alg=HS256 and sign a forged token using Apple's publicly available RSA public key as the HMAC secret, bypassing signature verification and impersonating any Apple-linked account.
Casky was already ahead
This CVE exploits attack patterns that Casky's 0matched 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.
CVE-2026-82454 represents a critical authentication bypass in Omnivore's Apple sign-in implementation, stemming from unsafe JWT token verification practices. The vulnerability exploits a common cryptographic pitfall: the decodeAppleToken function extracts the 'alg' parameter directly from untrusted JWT headers without validation, then passes it as the sole permitted algorithm to jwt.verify(). An attacker can forge valid tokens by setting alg=HS256 (HMAC-based) while signing with Apple's publicly available RSA public key as the HMAC secret—a technique known as algorithm confusion. This affects any deployment using the vulnerable Omnivore API version, potentially compromising user authentication and authorization across integrated applications that rely on Apple sign-in for account access.
While Casky.ai currently has zero mapped skills specifically for this CVE, practitioners using Claude AI with extended reasoning can detect the attack patterns underlying this vulnerability through reconnaissance and credential access analysis. The exploitation chain manifests as suspicious JWT token anomalies: unexpected algorithm fields in authentication headers, mismatches between declared and actual signature schemes, and successful authentication using public key material as symmetric secrets. Security teams should implement detection rules monitoring for algorithm confusion patterns (CWE-347), validate algorithm-key compatibility before verification, and enforce strict JWT validation libraries that prevent attacker-controlled algorithm selection. Practitioners would benefit from Casky skill development targeting secure cryptographic implementation patterns and JWT validation best practices to surface these logic flaws during code review and threat modeling exercises.
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 →
Casky has 0 skills that investigate the attack patterns behind CVE-2026-82454. Run one and get CVSS-scored findings in 3 minutes.
Run the skill that detects this →© 2026 Casky.AI, Inc. · AI Security Investigation