Critical Red Hat Quay Flaw Lets Attackers Read and Overwrite All Private Container Images in a Registry

Red Hat has disclosed two critical vulnerabilities in its Quay container registry that together let attackers read and tamper with every private image stored in an instance. The first, CVE-2026-123710 (CVSS 9.8), abuses a flaw in how Quay's key server validates token-signing keys, allowing an attacker—often without any account, since anonymous access is on by default—to plant a trusted, never-expiring signing key and forge tokens for pushing, pulling, or deleting images across all repositories, as well as wiping legitimate signing keys to break normal registry operations. The second, CVE-2026-110697 (CVSS 9.9), exploits an unescaped SQL LIKE pattern in the robot-account lookup logic, so a user who can create a namespace containing an underscore can harvest another tenant's robot account tokens and hijack their images, with no special privileges required. Red Hat credits its own engineer Dave O'Connor with finding both bugs, notes the first flaw has existed since Quay's earliest commits, and has shipped fixes across eight Quay 3 release branches while urging admins to restrict access to the /keys endpoint, audit logs for suspicious key-rotation events, limit who can create namespaces, and rotate robot tokens until patches are applied.