diff --git a/apps/victoria-metrics/values.yaml b/apps/victoria-metrics/values.yaml
index 28cd91d..96f463a 100644
--- a/apps/victoria-metrics/values.yaml
+++ b/apps/victoria-metrics/values.yaml
@@ -62,6 +62,17 @@ victoria-metrics-k8s-stack:
       limits:
         memory: 128Mi
 
+  # Talos binds the scheduler and controller-manager metrics endpoints to
+  # 127.0.0.1, and runs etcd as a host service rather than a Pod. Scraping them
+  # from vmagent would require exposing those control-plane ports on the LAN,
+  # so skip them; API server, kubelet, CoreDNS, and node metrics remain.
+  kubeScheduler:
+    enabled: false
+  kubeControllerManager:
+    enabled: false
+  kubeEtcd:
+    enabled: false
+
   # kube-state-metrics reports desired/current Kubernetes object state (for
   # example unavailable Deployment replicas), complementing host metrics.
   kube-state-metrics: