Skip to main content
Connect a Kubernetes cluster so Myrmex can inspect your workloads, review RBAC, audit cluster hardening, and help you troubleshoot. Myrmex reaches the cluster’s API server through a Collector using a service account token — everything is read in real time, on demand, and nothing from your cluster is stored on the Myrmex side.

What Myrmex Can Do

Inspect workloads

List and describe pods, deployments, services, and nodes — read-only diagnostics.

Cluster posture

Run Security Posture hardening checks across the cluster.

RBAC review

Review roles, bindings, and service accounts for over-permissive access.

Troubleshoot & act

Investigate failing workloads and, on approval, apply manifest changes.

Before You Start

  • A Collector that can reach the cluster’s API server over HTTPS. See Deploying the agent.
  • A service account in the cluster with a token and a least-privilege role — the built-in view ClusterRole is enough for analysis; grant write access only for remediation.
  • The cluster’s API server URL and CA certificate.

Step 1 — Create the Credential in Kubernetes

1

Create a service account

Create a dedicated service account, for example kubectl create serviceaccount myrmex -n kube-system.
2

Bind a least-privilege role

Bind the built-in view ClusterRole (or a scoped Role) to the service account for read-only access. Grant a write-capable role only when you want Myrmex to remediate.
3

Create a token

Issue a token with kubectl create token myrmex -n kube-system (or read it from a bound Secret). Copy the token value.
4

Get the API server and CA

From your kubeconfig, copy the cluster server URL and the certificate-authority-data value.
Keep the service account bound to view for analysis, and only bind a write role once you want the agents to apply changes.

Step 2 — Add the Integration in Myrmex

From the Directory, choose Add Integration → Kubernetes, then fill in:

Connect

Click Connect to validate the connection and finish. The cluster then appears under Environment → Integrations, and you can start asking the AI about it in the Workspace.
Prefer least privilege: a service account bound to the view ClusterRole covers analysis, and you can rotate its token at any time.