Red Hat Quay — Critical Signing Key Injection and Credential Theft Vulnerabilities (CVE-2026-123710)
Publication date: November 12, 2026
Category: Vulnerability / Cloud Security
Introduction
Security researchers have disclosed two extreme-severity vulnerabilities affecting Red Hat Quay, a prominent enterprise container registry platform heavily deployed across standalone datacenters and cloud-native environments, including Red Hat OpenShift infrastructures. Tracked as CVE-2026-123710 and CVE-2026-110697, these flaws empower malicious actors to subvert the registry’s cryptographic trust model, read confidential data, overwrite private container images, and execute cross-tenant credential theft. Red Hat has released patched builds across multiple Quay 3 and Quay 5 branches, urging administrators to apply updates promptly and implement perimeter mitigations.
What is Red Hat Quay? (General Analysis)
Red Hat Quay is a high-availability private container registry designed to store, manage, and secure container images within enterprise environments. It functions as a critical backbone for continuous integration and continuous deployment (CI/CD) pipelines, supplying trusted artifacts to container orchestration platforms such as Red Hat OpenShift.
The discovered vulnerabilities dismantle the foundational assumptions of isolation and cryptographic integrity inherent to the platform:
- CVE-2026-123710 (CVSS 9.8 - CRITICAL): Mapped to weakness CWE-1287 (Improper Validation of Specified Type of Input / logic flaw in key rotation). It enables any client—operating anonymously if anonymous access is enabled, or with a standard user account otherwise—to obtain an arbitrary registry token, present it to the service key rotation endpoint, and install an unauthorized signing key with no expiration date. Consequently, any subsequent token signed with that injected key is universally accepted across the registry.
- CVE-2026-110697 (CVSS 9.9 - CRITICAL): Mapped to weakness CWE-89 (Improper Neutralization of Special Elements used in an SQL Command / SQL Injection). Located within the robot account listing function, it dynamically constructs an SQL
LIKEpattern using user-supplied namespace names without escaping wildcard metacharacters (such as the underscore_, which matches any single character in SQLLIKE). This allows a low-privileged user creating a specific namespace to match adjacent tenant namespaces and retrieve their robot accounts along with decrypted authentication tokens.
How Does It Work? (Technical Analysis)
An examination of the operational behavior underlying these vulnerabilities exposes architectural flaws in request validation and dynamic query construction:
-
Exploitation Flow for Key Injection (CVE-2026-123710):
- Quay relies on a set of trusted cryptographic keys to sign and validate access tokens required for image push and pull operations.
- The Quay key server verifies that incoming key replacement requests carry a token signed by a trusted key matching expected audience and issuer parameters. However, it fails to validate whether the token was specifically issued for a legitimate key rotation workflow.
- An attacker queries the
/v2/authendpoint to obtain an anonymous token with a self-selected audience. Quay signs this token using a trusted key. - The attacker submits this token to the key rotation endpoint along with their own public key configured with an indefinite lifespan.
- Once the malicious key is successfully registered, the attacker can sign fabricated tokens granting full read and write access to any private repository, maintaining persistence across service restarts until manually purged.
-
Exploitation Flow for Robot Account Theft (CVE-2026-110697):
- The internal code responsible for listing robot accounts ingests a user-provided namespace name and constructs an SQL
LIKEquery pattern. - Due to insufficient sanitization of wildcard characters, an attacker registers a namespace containing an underscore character (
_). - During database execution, the underscore acts as a single-character wildcard, causing the query to overlap with other tenants’ namespaces and return their associated robot accounts.
- The attacker successfully harvests decrypted authentication tokens, enabling unauthorized image operations and potential code injection into integrated OpenShift CI/CD pipelines.
- The internal code responsible for listing robot accounts ingests a user-provided namespace name and constructs an SQL
Affected Systems / Environments
The vulnerabilities affect multiple deployments within the Quay ecosystem:
- All versions of Quay 3 and Quay 5.
- Standalone installations.
- Deployments managed by the Quay Operator on Red Hat OpenShift.
- Mirror registries for Red Hat OpenShift.
- Internal registries reachable by untrusted clients.
| CVE | Category (CWE) | Impact | CVSS | Vector (summary) |
|---|---|---|---|---|
| CVE-2026-123710 | CWE-1287 | Global read/write access to private images and tag tampering | 9.8 (CRITICAL) | AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H |
| CVE-2026-110697 | CWE-89 | Cross-tenant credential theft and pipeline injection | 9.9 (CRITICAL) | AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H |
Mitigation and Detection
Remediation
- Software Updates: Immediately deploy the official patched builds released by Red Hat for the vulnerable Quay 3 and Quay 5 branches. Updating the application is the sole comprehensive and definitive fix.
- Perimeter Restriction: At the load balancer or reverse proxy level, block external access to the
/keysendpoint, restricting traffic exclusively to trusted internal components. - Namespace Policies: Temporarily restrict the creation of organizations or namespaces containing wildcard characters such as underscores (
_).
Detection
- Log Auditing: Inspect audit logs for
service_key_rotateandservice_key_deleteevents where theexpiration_dateparameter is evaluated as null. - Anomaly Monitoring: Monitor registry activity for unusual pull or push operations that deviate from standard robot account behavioral baselines.
Wrapping Up
The discovery of CVE-2026-123710 and CVE-2026-110697 in Red Hat Quay underscores severe risks to containerized software supply chains. By permitting the unauthorized injection of signing keys and mass extraction of robot credentials via SQL handling flaws, threat actors can quietly compromise the operational integrity of entire cloud-native environments. Immediate patching and strict perimeter hardening are essential countermeasures to mitigate exposure.
References
- Khandelwal, S. (2026). Critical Red Hat Quay Flaw Lets Attackers Read and Overwrite All Private Container Images in a Registry. The Hacker News. https://thehackernews.com/2026/11/critical-red-hat-quay-flaw-lets.html
- Red Hat Bugzilla. (2026). Bug 2568072 - Red Hat Quay vulnerability CVE-2026-123710. https://bugzilla.redhat.com/show_bug.cgi?id=2568072
- Red Hat Bugzilla. (2026). Bug 2552237 - Red Hat Quay vulnerability CVE-2026-110697. https://bugzilla.redhat.com/show_bug.cgi?id=2552237
- Red Hat Security. (2026). CVE-2026-123710 Advisory and Details. https://access.redhat.com/security/cve/cve-2026-123710
- Red Hat Security. (2026). CVE-2026-110697 Advisory and Details. https://access.redhat.com/security/cve/cve-2026-110697
- Red Hat Support. (2026). Key remediation and mitigation steps for Red Hat Quay. https://access.redhat.com/solutions/7149923
- Red Hat Support. (2026). Mitigation guidance for cross-tenant robot token flaws. https://access.redhat.com/solutions/7149592
- Red Hat Quay Upstream Project. (2026). Upstream fix commit for key rotation and search logic. https://github.com/quay/quay/commit/094cfb024f01