The cluster already knows why the pod died. Polylane reads the events, correlates the rollout, and gives you the answer.
Any cluster that speaks the Kubernetes API: EKS, GKE, AKS, or the one under your desk. One read-only ServiceAccount and it's in the graph.
Join the waitlistEvery supported resource.
The Kubernetes resource types that show up in your graph.
Namespace Node Deployment StatefulSet DaemonSet ReplicaSet Job CronJob Pod Service Ingress ConfigMap Secret Service Account Role Role Binding Cluster Role Cluster Role Binding Persistent Volume Persistent Volume Claim Storage Class Network Policy Horizontal Pod Autoscaler Pod Disruption Budget The cluster knows. Now someone's listening.
Agents watch the workloads and investigate failures as they happen.
The full cluster topology
Deployments, StatefulSets, DaemonSets, pods, services, ingresses, CronJobs, volumes, autoscalers, roles and bindings: 24 resource types land in one graph with their edges and history, alongside the clouds the cluster runs on.
Agents inspect the cluster directly
They inspect pods, read logs and events, and map the topology on a cadence, judging what they see against how each workload normally behaves. No dashboards in the loop.
The failures Kubernetes hides well
Crashloops, OOM kills, pods stuck pending, rollouts that never converge. The cluster knows; nobody's watching. Polylane investigates and writes up what happened.
What changed before it broke
Each sync records what appeared, changed, or disappeared in the cluster. When a workload starts failing right after a rollout, the investigation starts from that rollout.
get, list, watch, and nothing else
Connect with a kubeconfig or a read-only ServiceAccount token. Polylane only makes read-only Kubernetes API calls, credentials are encrypted before they're stored, the agent never sees the token, and you can scope it to a single namespace.
Investigations that end in a PR
When the root cause is a line of code or a manifest, the fix lands on the repo behind the workload, evidence attached, ready for your review.
Questions.
How does Polylane connect to my Kubernetes cluster?
Two ways: paste a kubeconfig (Polylane reads the API server URL, CA certificate, and ServiceAccount bearer token from the current context), or enter the API server URL and a token manually. Create a read-only ServiceAccount with get, list, and watch, and that's all it needs. Credentials are encrypted before they're stored, and the agent never sees the token.
Does it work with EKS, GKE, or my own clusters?
Any cluster that speaks the Kubernetes API works: EKS, GKE, AKS, or bare metal. If you connect an AWS account, EKS clusters, node groups, and Fargate profiles also show up in the same graph automatically.
What Kubernetes resources does Polylane sync?
24 resource types: namespaces, nodes, Deployments, StatefulSets, DaemonSets, ReplicaSets, Jobs, CronJobs, pods, services, ingresses, ConfigMaps, secrets, ServiceAccounts, RBAC roles and bindings, persistent volumes and claims, storage classes, network policies, autoscalers, and pod disruption budgets. The full list is on this page.
Can Polylane change my cluster?
No. Polylane only makes read-only Kubernetes API calls; the ServiceAccount you create needs nothing beyond get, list, and watch. Fixes land as pull requests on the repos behind your workloads, for you to review and merge.
Does it work with my observability provider?
Yes. Datadog, Sentry, Honeycomb, and Axiom join the graph as first-class sources, and their alerts get triaged by an agent the moment they fire.
What does it cost?
Pricing is public and self-serve: a free plan, paid plans from $80 a month, and prepaid autofix bundles that never expire. Every plan and limit is published at polylane.com/pricing.