ktop is an Apache-2.0 licensed, Go-based terminal tool that shows live metrics for nodes, pods, and containers in a Kubernetes cluster, built for operators and developers who want top-style cluster visibility without installing anything inside the cluster.

What it is

ktop is a single-binary command-line tool that follows the tradition of Unix/Linux top. It displays useful metrics information about nodes, pods, and other workload resources running in a Kubernetes cluster, and it ships as a kubectl plugin, a Homebrew formula, a go install target, and a downloadable binary on GitHub Releases. It lives in the Kubernetes ecosystem, speaks to the cluster through the Kubernetes API using a local kubeconfig, and needs no agents or server-side components.

The concrete problem it solves is the switching cost of reading cluster state. kubectl top returns a single snapshot of node and pod CPU and memory, while kubectl logs streams one container at a time; correlating those outputs across namespaces and contexts means running several commands and holding the results in your head. ktop replaces that with one continuously updated view that drills down from a cluster overview to a node, then to a pod, then to container logs, and it connects anywhere kubectl connects, so no direct node access is required.

Key capabilities

• Real-time monitoring of nodes, pods, and containers in a continuously updated view.
• Drill-down navigation from cluster overview to node to pod to container logs.
• Zero cluster installation: runs locally from a kubeconfig, with no agents or server-side components.
• Works anywhere kubectl works, because it connects through the Kubernetes API rather than to nodes directly.
• Streaming container logs with filtering, timestamps, line wrap, and a full-screen mode.
• Multiple metrics sources: Prometheus for enhanced metrics via --metrics-source=prometheus, Metrics-Server, or a no-metrics fallback.
• Graceful degradation that still shows resource requests and limits when no metrics infrastructure is present.
• Context and namespace scoping through --context my-cluster and --namespace production.

Who uses it and how

• Platform and operations engineers running multi-namespace clusters use --namespace to scope the view to a single team or application instead of the whole cluster.
• Engineers who work across several clusters use --context to switch targets without rewriting kubeconfig or re-running discovery commands.
• Teams that already run Prometheus point ktop at it with --metrics-source=prometheus for richer metrics, while teams without a metrics stack fall back to Metrics-Server or to the requests-and-limits view.
• Debugging workflows benefit from staying in one terminal: a user navigates to a pod and reads streaming container logs with timestamps and filtering rather than opening a second command.
• Managed or sandboxed clusters where node SSH is unavailable remain workable, because the API-only connection needs nothing beyond a valid kubeconfig.

Getting started

Install as a krew plugin with kubectl krew install ktop and run it as kubectl ktop, or use brew tap vladimirvivien/oss-tools followed by brew install ktop. Alternatives are go install github.com/vladimirvivien/ktop@latest or a binary from GitHub Releases.

How it compares

The supplied facts name no other tools in this registry that ktop competes with or replaces, so it stands alone here. Its reference point is the Unix/Linux top family rather than a named peer product, and it is positioned as an extension of the kubectl workflow rather than a replacement for it.

When to use it — and when not to

A self-hoster needs only a valid kubeconfig, taken from $KUBECONFIG or ~/.kube/config, and optionally a Prometheus or Metrics-Server endpoint for richer data. Do not pick ktop if you need retained history, dashboards, or alerting, since it is a live terminal view and the facts describe no persistence layer. Documentation lives mainly on the project website, and the README itself is brief, so expect to consult the external site before advanced use.

kubectl krew install ktop
kubectl ktop

brew tap vladimirvivien/oss-tools
brew install ktop

go install github.com/vladimirvivien/ktop@latest

# Start ktop with default settings
ktop

# Use a specific context
ktop --context my-cluster

# Filter to a namespace
ktop --namespace production

# Use Prometheus for enhanced metrics
ktop --metrics-source=prometheus