CVE-2026-10059 Overview
CVE-2026-10059 is a privilege escalation flaw in the Multicluster Engine for Kubernetes ClusterCurator controller. A tenant administrator holding only namespace-scoped privileges can create a namespaced ClusterCurator object. That action inadvertently allows the tenant administrator to mint a token for a ServiceAccount with cluster-wide administrative authority. The result is full control over the cluster from a scoped tenant role. The weakness maps to CWE-266: Incorrect Privilege Assignment.
Critical Impact
A namespace-scoped tenant administrator can escalate to full cluster-admin, breaking multi-tenant isolation and compromising confidentiality, integrity, and availability of every workload on the cluster.
Affected Products
- Red Hat Advanced Cluster Management / Multicluster Engine for Kubernetes
- ClusterCurator controller component
- Deployments exposing namespaced ClusterCurator resources to tenant administrators
Discovery Timeline
- 2026-08-05 - CVE-2026-10059 published to NVD
- 2026-08-05 - Last updated in NVD database
Technical Details for CVE-2026-10059
Vulnerability Analysis
The ClusterCurator controller in the Multicluster Engine automates lifecycle operations such as provisioning, upgrading, and destroying managed clusters. To perform those operations, it associates each ClusterCurator with a ServiceAccount that carries broad, cluster-wide permissions. The controller trusts the namespaced ClusterCurator resource to declare which ServiceAccount to act on. A tenant administrator who only holds create/update rights within a single namespace can therefore instruct the controller to obtain a token for a privileged ServiceAccount. Once minted, the token grants cluster-admin equivalent access outside the tenant boundary.
Root Cause
The controller performs privileged operations on behalf of a low-privileged requester without verifying that the requester is authorized to consume the referenced ServiceAccount. This is a classic incorrect privilege assignment pattern [CWE-266]: authority flows from the controller's own credentials rather than the caller's, and the caller's namespace scope is not enforced against the target ServiceAccount.
Attack Vector
An authenticated tenant administrator with namespace-scoped rights creates or modifies a ClusterCurator custom resource in a namespace they control. The resource references a privileged ServiceAccount used by the controller for cluster lifecycle actions. The controller reconciles the object and mints or exposes a token bound to that ServiceAccount. The attacker retrieves the token and issues cluster-scoped API calls, including reading secrets across namespaces, modifying workloads, and creating additional persistent access. No user interaction and no additional network position is required beyond authenticated API access.
No public proof-of-concept is currently listed. Refer to the Red Hat CVE-2026-10059 Advisory and Red Hat Bugzilla Report #2483187 for vendor technical detail.
Detection Methods for CVE-2026-10059
Indicators of Compromise
- Creation or modification of ClusterCurator custom resources by tenant-scoped identities that do not normally manage cluster lifecycle.
- TokenRequest or ServiceAccount token issuance events for privileged lifecycle ServiceAccounts originating from reconciliation of a namespaced ClusterCurator.
- Cluster-scoped API calls (secret reads, ClusterRoleBinding writes, node operations) authenticated as a lifecycle ServiceAccount shortly after ClusterCurator activity.
Detection Strategies
- Audit the Kubernetes API server log for create/update verbs on clustercurators.cluster.open-cluster-management.io correlated with subsequent serviceaccounts/token requests.
- Alert when a namespaced principal triggers reconciliation that results in cluster-scoped resource changes attributed to a controller ServiceAccount.
- Baseline which identities legitimately author ClusterCurator resources and flag deviations.
Monitoring Recommendations
- Forward Kubernetes audit logs and controller logs to a central analytics platform such as Singularity Data Lake for correlation across tenants.
- Monitor for anomalous use of high-privilege ServiceAccount tokens, especially API calls sourced from tenant namespaces.
- Track RoleBinding and ClusterRoleBinding changes made by service accounts associated with the Multicluster Engine.
How to Mitigate CVE-2026-10059
Immediate Actions Required
- Apply the fixed Multicluster Engine build referenced in the Red Hat CVE-2026-10059 Advisory as soon as it is available for your channel.
- Inventory every namespace where tenant administrators can create ClusterCurator objects and temporarily restrict that permission.
- Rotate tokens and credentials for any ServiceAccount used by the ClusterCurator controller.
- Review audit logs for unexpected ClusterCurator creations and privileged API calls since the component was deployed.
Patch Information
Red Hat tracks remediation in Bugzilla #2483187 and the Red Hat CVE-2026-10059 Advisory. Upgrade Multicluster Engine and any dependent Advanced Cluster Management deployments to the vendor-fixed release. Confirm that the controller no longer mints tokens for privileged ServiceAccounts based on tenant-supplied references.
Workarounds
- Remove tenant administrators from RBAC roles that permit create, update, or patch on ClusterCurator resources until patched.
- Constrain the ClusterCurator controller ServiceAccount to the minimum permissions required, and separate lifecycle ServiceAccounts per managed cluster.
- Enforce an admission policy (for example, OPA/Gatekeeper or Kyverno) that rejects ClusterCurator objects referencing ServiceAccounts outside an approved allowlist.
- Isolate the Multicluster Engine namespace so that only cluster operators, not tenants, hold write access to lifecycle custom resources.
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
