Multicluster Engine CVE-2026-66794

Exploitation requires network reachability to the MCE hub's user-facing cluster-proxy-addon route and the ability to craft URL path segments that select a target service and managed cluster; per the CVSS vector (PR:N/UI:N) it is unauthenticated and needs no user interaction. The concrete prerequisite is that the proxy route is exposed to the attacker and that the cluster-proxy-addon is deployed and managing at least one downstream cluster with internal services worth reaching. The primary limiting factor is exposure of that route: deployments that keep the proxy route on an internal-only management network, behind a VPN, or firewalled off are not remotely exploitable by external attackers, whereas any hub with the route reachable from an untrusted network meets all conditions against effectively default configurations.

The signals point to a genuine high-priority issue rather than an inflated score: the CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:L/A:N, 9.3) describes remote, low-complexity, unauthenticated exploitation with a scope change reflecting cross-cluster reach, which is exactly what the SSRF/auth-bypass description supports - the metrics and the narrative are consistent. No EPSS score was provided and the CVE is not in CISA KEV, and no proof-of-concept was supplied, so there is no evidence of active or automated exploitation yet; the risk here is driven by exposure and impact, not by observed attacks. The most important limiting factor is reachability: exploitation requires network access to the user-facing proxy route, so hubs whose route is exposed to untrusted networks are at materially higher risk than those restricted to internal management networks, and a mid-sized operator running an internet-adjacent MCE hub should treat this as urgent while an air-gapped or tightly network-segmented deployment has a meaningfully smaller attack surface. Because MCE governs many clusters at once, the blast radius of a single successful request is large, which justifies prioritizing remediation even in the absence of a public exploit; note that EPSS and KEV data are simply absent here rather than confirming low risk.

An attacker who can reach the internet-exposed MCE hub proxy route sends an HTTP request with crafted URL path segments that name an internal service on a managed cluster. The proxy bypasses authorization and forwards the request, returning data from an internal API or service (for example a cluster's internal metadata or admin endpoint) to the unauthenticated attacker. No proof-of-concept is publicly identified at time of analysis, but the low attack complexity (AC:L) and lack of required privileges (PR:N) make weaponization straightforward for a motivated attacker with route access.