CVE-2026-44935 CrowdStrike LogScale · LogScale

Detect Rancher Fleet Cross-Namespace Secret Disclosure via Unvalidated valuesFrom References in CrowdStrike LogScale

CVE-2026-44935 is a critical (CVSS 9.9) authorization bypass vulnerability in Rancher Fleet's Helm Deployer affecting versions 0.12.0-0.12.14, 0.13.0-0.13.10, 0.14.0-0.14.5, and 0.15.0-0.15.1. The Helm Deployer fails to validate namespace boundaries when resolving `valuesFrom` references in GitRepo or Bundle resources, allowing an attacker with access to one namespace to craft a GitRepo or Bundle that references Secrets or ConfigMaps from arbitrary namespaces including cluster-scoped secrets. This constitutes an incorrect authorization check (CWE-863) that can expose credentials, API keys, and sensitive configuration from namespaces the attacker should not have access to. A public PoC is available.

MITRE ATT&CK

Tactic
Credential Access Discovery

LogScale Detection Query

CrowdStrike LogScale (LogScale)
cql
#event_simpleName=KubernetesAuditEvent
| verb IN ("get", "list", "watch")
| objectRefResource IN ("secrets", "configmaps")
| userUsername = /system:serviceaccount:.*fleet.*/i
| responseStatusCode = 200
| eval source_namespace=replace(userUsername, /system:serviceaccount:([^:]+):.*/, "\\1")
| where objectRefNamespace != source_namespace AND isNotNull(objectRefNamespace)
| stats count() AS cross_ns_count, values(objectRefName) AS secret_names, values(objectRefNamespace) AS target_namespaces BY userUsername, source_namespace, ClusterName
| where cross_ns_count > 0
| sort -cross_ns_count
critical severity medium confidence

CrowdStrike Falcon LogScale (CQL) query identifying Rancher Fleet service account cross-namespace secret reads consistent with CVE-2026-44935 exploitation via unvalidated valuesFrom references.

Data Sources

CrowdStrike Falcon Kubernetes Admission ControllerKubernetes audit log ingestion via Falcon LogScale

Required Tables

KubernetesAuditEvent

False Positives & Tuning

  • Fleet workloads with CrowdStrike exclusions or namespace-level monitoring gaps
  • Kubernetes clusters where audit logging verbosity does not capture secret GET operations
  • Service accounts with 'fleet' in their name that are not part of Rancher Fleet

Other platforms for CVE-2026-44935


Testing Methodology

Validate this detection against 4 adversary techniques from Atomic Red Team. Each test below lists the behaviour to exercise and the telemetry you should expect to see. Executable commands and cleanup steps are available with Pro.

  1. Test 1Simulate Fleet valuesFrom Cross-Namespace Secret Read

    Expected signal: Kubernetes audit log entry: verb=get, resource=secrets, objectRef.name=victim-secret, objectRef.namespace=fleet-test-victim, user.username=system:serviceaccount:fleet-test-attacker:fleet-*, responseStatus.code=200

  2. Test 2Enumerate Existing GitRepo valuesFrom Cross-Namespace References

    Expected signal: Script output listing cross-namespace references; Kubernetes API audit logs for GET gitrepos and GET bundles at cluster scope

  3. Test 3Fleet Service Account RBAC Permission Audit for Secret Access

    Expected signal: kubectl auth can-i commands generate Kubernetes audit log entries for SubjectAccessReview API calls; RBAC audit output identifies over-privileged Fleet service accounts

  4. Test 4Monitor Fleet Controller Live for Cross-Namespace Secret Access Events

    Expected signal: Real-time stdout alerts for each cross-namespace secret or configmap read event matching Fleet service account patterns

Unlock Pro Content

Get the full detection package for CVE-2026-44935 including response playbook, investigation guide, and atomic red team tests.

Response PlaybookInvestigation GuideHunting QueriesAtomic Red Team TestsTuning Guidance

Related Detections