Red Hat OpenShift WMCO CVE-2026-54100

Exploitation requires the attacker to be on the same adjacent network segment as either the WMCO controller pod or the target Windows worker node and to perform an active MITM (ARP/NDP spoofing, rogue DHCP, or upstream router compromise) during the narrow window when WMCO initiates the SSH bootstrap to a new or reconfigured Windows node. It is limited to clusters that actually deploy Windows worker nodes via the Windows Machine Config Operator (Red Hat OpenShift for Windows Containers); Linux-only OpenShift clusters are not affected. No authentication to OpenShift is required (PR:N), but the attacker must already hold a privileged L2/L3 position on the machine network, and there is no user-interaction requirement.

The CVSS:3.1 vector AV:A/AC:H/PR:N/UI:N/S:C/C:H/I:H/A:H yields 8.3 (High), driven largely by the scope change (S:C) - stealing bootstrap credentials lets the attacker pivot from a single node to cluster-level Windows workloads. However, two strong limiters apply: AV:A constrains the attacker to the adjacent network (typically the cluster/management VLAN, not the public internet), and AC:H reflects the need to win an SSH MITM race during the brief bootstrap window. There is no EPSS, KEV, or SSVC data supplied, no POC has been published, and exploitation requires already holding a privileged network position; the realistic risk is therefore lower than the 8.3 base score for most operators but meaningful for multi-tenant or shared-network clusters where lateral attackers may already sit on the node subnet. CVSS attack vector and the description are consistent (both adjacent/MITM), so no contradiction to flag.

An attacker who has gained a foothold on the OpenShift machine/management subnet (for example via a compromised infrastructure node or a misconfigured tenant workload sharing the VLAN) ARP-spoofs or BGP-hijacks the IP of a newly provisioned Windows node and answers WMCO's SSH connection with their own server. Because WMCO does not verify the host key, the handshake succeeds and WMCO transfers the WICD binary and kubelet bootstrap kubeconfig to the attacker, who then uses those credentials to register a rogue Windows node identity and read or tamper with Windows workloads in the cluster. No public PoC has been observed.