Ask in plain English
Tag any pod, node, or namespace with @ and ask. The assistant answers from live cluster state — never a guess.
Kubevue turns your kubeconfig into a fast, navigable map — then lets you tag any resource with @ and ask in plain English. Read-only by default; it answers only from what your cluster actually reports.
Scroll to try itpostgres-0 has restarted 6 times in the past hour. The previous container exited with OOMKilled (exit code 137) — memory hit the 512Mi limit during a checkpoint. Recent events show Back-off restarting failed container.
Next step — raise the container memory limit (or lower shared_buffers), then watch the next checkpoint complete in the pod logs.
Product demo — example cluster. Kubevue shows only what your cluster actually reports.
Build your own view
Add, drag, and resize widgets anywhere on a cluster’s overview — stat cards, workload health, recent events, or a live Grafana panel pulled straight from your own dashboards.
Across every cluster
Build a global dashboard that pulls widgets from many clusters at once — a cross-fleet “needs attention” list, or interleaved logs from two services talking to each other across regions.
Read-only, by design
Tag any pod, node, or namespace with @ and ask. The assistant answers from live cluster state — never a guess.
Namespaces and workloads laid out as an explorable “city,” with deterministic layout and pan / zoom.
Full dashboards in-app via a sandboxed webview, or individual panels pinned as live tiles. Per-cluster URL and key.
Automatic kubeconfig discovery with EKS-aware metadata — provider and region — surfaced up front.
Keeps the last good snapshot when a refresh fails, with clear disconnected-state UX and retry actions.
Every value comes from the live cluster. Nothing synthesized, estimated, or faked — if it isn’t real, it isn’t shown.