Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Admission controllers intercept requests to the Kubernetes API after authentication and authorisation, and can mutate or reject them—making them the enforcement point for cluster policy.

What Are Kubernetes Admission Controllers?
An admission controller is a piece of code that intercepts a request to the Kubernetes API server after the request has been authenticated and authorised but before the object is persisted. It can modify the object, reject it, or let it through. That position—last gate before anything exists in the cluster—makes admission control the natural place to enforce security policy, because a rule applied here cannot be bypassed by a user who holds the right RBAC permissions.
Admission runs in two phases. Mutating controllers go first and may change the object: injecting a sidecar, adding a default security context, setting resource limits, or applying labels. Validating controllers run afterwards and may only accept or reject, which means they see the final version of the object including any mutations.
The ordering matters when you design policy. A validating rule that rejects containers without a read-only root filesystem will approve a pod whose mutating webhook has just added one. Teams sometimes rely on that deliberately—mutate to a safe default, then validate that nothing slipped through—and sometimes discover it by accident when a rule they trusted turns out never to fire.
Kubernetes ships with several dozen compiled-in controllers, including NamespaceLifecycle, LimitRanger, ServiceAccount, and ResourceQuota. The one that matters most for security is Pod Security Admission, which became stable in Kubernetes 1.25 when PodSecurityPolicy was removed.
Pod Security Admission applies one of three profiles per namespace—privileged, baseline, or restricted—in one of three modes: enforce, audit, or warn. The restricted profile blocks privileged containers, host namespaces, host path mounts, added capabilities beyond a tiny set, and containers running as root. Applying restricted in enforce mode to application namespaces closes most of the routes described in our entry on the container escape, and it costs nothing beyond a namespace label. Start in warn and audit modes to see what would break.
Pod Security Admission is deliberately narrow and not configurable beyond its three profiles. For organisation-specific rules you need a policy engine running as a webhook:
Typical rules worth writing: require images from approved registries and verify their signatures, forbid the latest tag, require resource requests and limits, require specific labels for ownership, block services of type LoadBalancer in certain namespaces, and enforce that every pod has a non-default service account.
An external webhook is a dependency of your cluster’s control plane, and the failurePolicy field decides what happens when it is unavailable. Set to Fail, an outage of the policy engine stops all matching API requests—including deployments, scaling, and sometimes the recovery of the policy engine itself. Set to Ignore, the cluster keeps working and your policy silently stops being enforced.
Neither default is right everywhere. Use Fail for security-critical policies, run the webhook with multiple replicas across nodes and a pod disruption budget, exclude the kube-system namespace and the policy engine’s own namespace from its scope, and keep timeouts short. Then test the failure explicitly rather than discovering the behaviour during an incident.
Admission runs once, at write time. It cannot see what a container does after it starts, so a compliant pod that downloads and executes something unexpected is entirely outside its view—that is the job of runtime detection and a cloud workload protection platform. It also does not affect objects already in the cluster, so apply a background audit scan when you introduce a new policy, and pair it with network policies for traffic control between the pods it admits.
At Privia Security, we assess cluster policy and attempt to bypass it through our cloud security services.
Pod Security Admission, which became stable in Kubernetes 1.25 when PodSecurityPolicy was removed. It applies one of three fixed profiles—privileged, baseline, or restricted—per namespace, in enforce, audit, or warn mode, controlled by namespace labels. It is simpler but far less flexible, so organisations needing custom rules pair it with a policy engine such as Kyverno or OPA Gatekeeper.
Use Fail for policies that must not be bypassed, and accept that an outage of the webhook will block matching API requests. Use Ignore only for advisory or non-security rules. Whichever you choose, run the webhook highly available across nodes, exclude kube-system and the engine’s own namespace from its scope, keep timeouts short, and test the failure path deliberately.
Yes, and it is one of the strongest uses of it. Kyverno and Gatekeeper can both verify Sigstore Cosign or Notation signatures and reject unsigned or wrongly signed images, which prevents a tampered or unreviewed image from being scheduled. Combine it with a registry allowlist and digest pinning so the policy cannot be satisfied by an image from somewhere unexpected.
No. Admission runs only when an object is created or updated, so existing workloads are unaffected until they are next changed or rescheduled. Introduce a new policy in audit or warn mode first, run a background scan to list the existing resources that violate it, remediate those through normal change control, and only then switch to enforce.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.