Release notes

Kubby now finds Claude Code, Codex and Gemini CLI even when they are not on PATH, and sees a PATH change without a restart, so a tool you installed yourself no longer shows as not installed. If an installation started from Kubby fails, its window stays open with the message instead of closing. Claude Code installed only in WSL is now detected and shown in Preferences; chats still need the Windows install.

Live lists now handle network hiccups the way Lens does: when a watch connection drops in the middle, Kubby quietly carries on where it left off instead of showing "Could not connect to the cluster". A real connection problem is still shown, now with its technical cause, and clears by itself once the cluster answers again. Pods with native sidecar containers (for example Istio) show Running instead of Init:1/2, Completed pods are green like running ones, and ConfigMap values no longer sit on a gray band.

Lists load much faster on large clusters and slow networks: Kubby now asks the API server for compressed lists, loads each kind once and shares it between the Overview, the lists and the health checks, and talks HTTP/2 like kubectl, so a cluster needs about 20 connections instead of 60. On a cluster with 800 pods behind a slow link, the Overview fills in about 2 seconds instead of 15 and Pods in under half a second. After a VPN switch or a proxy restart, lists come back to life within a minute instead of five.

The new Overview: nodes, pods, warnings of the last hour and the cluster's CPU and memory at the top; pick a node for its CPU and memory over the last hour, its disk, pods and versions. The node status table shows every node's CPU, memory and disk, next to the latest events. Disk usage comes from the kubelet (needs the nodes/proxy permission). The Nodes list and its details have a new layout; capacity and allocatable labels in the metrics chart no longer overlap. Large lists: more than 300 ticked rows stay in bulk actions, the Events list says when it keeps only the newest 2,000, look-ahead and other regular expressions work in the filter, the CSV export no longer skips rows while the list changes, and 5,000 pods scroll more smoothly.

Lists rebuilt for large clusters: Pods and Events are filtered and sorted in the background and only the rows on screen reach the window; a list of 50,000 pods opens and scrolls smoothly. The first list arrives in parts, live updates place only the rows that changed, and Kubby sends fewer updates while the window is busy. Watches nobody uses close when the limit is reached across 50 clusters and reconnect when you come back. A smaller installer.

Pods with several containers no longer overlap in the pod list: the Image column shows the first image and how many more there are (+N). Hover the row to see them all; every container's image is in the pod's details.

Kubby is now free: every feature, no licence, no trial, no account. New installer like Lens's: install for yourself without administrator rights, or for everyone on the computer. A start-up panel while Kubby loads; closing the window keeps Kubby in the notification area (quit from its menu or File → Exit). Kubby finds updates on kubby.tech by itself and installs them after your OK; every download is checked against Kubby's signature. Security fixes: secrets in logs and Ask Kubby are masked in every naming style, a pasted kubeconfig can no longer run a program, the node shell image is pinned by digest. New user guide in English and Turkish.

Cluster health card on the Overview: a grade from A to F, a score out of 100, scores per area (workloads, pods, nodes, network, storage, security), a 30-day trend and what to fix first. The Cluster health page: findings by rule, scores per namespace, accept a finding with a reason. System namespaces and allowed registries in Preferences.

Clusters without a log store or Prometheus get a setup offer: Loki with Alloy (PVC or S3) or a small Prometheus. Before anything is installed you see the CPU, memory and disk it uses on your nodes, the rights it needs and the images; private registries are supported. Installed after your approval, audited and connected to Kubby.

Logs → History: past logs from Loki, Elasticsearch and OpenSearch found in the cluster or added centrally in Preferences, including the logs of deleted pods and of the nodes (kubelet, containerd). A central Prometheus too. Passwords stay in Kubernetes Secrets you create; Kubby keeps only the reference.

The new Logs page: the logs of every pod and container of several workloads, labels or namespaces in one live timeline, with pod starts, kills and events between the lines. Log volume, levels, the busiest pods, restarts and a jump to the crash. Node, zone and time range filters.

Standard and Plus plans; Plus adds Monitoring, Argo CD, notifications, Ask Kubby and the audit log. A 14-day trial with every feature, one computer per key, a signed offline licence and updates from inside the app. Since 28 October 2026 Kubby is free for everyone, with every feature; the plans were removed.

Monitoring: node health, capacity planning, workloads, control plane, certificate expiry and alerts from the cluster's own Prometheus. Node cards, details that open on click and a table with adjustable columns.

Argo CD control centre: several Argo CD instances and clusters, an overview with charts, resource tree, diff, history, rollback and a "why Degraded" explanation. Ingress links and Gateway API templates.

Notifications: security patches and new versions for the components in your cluster, plus DevOps news.

Kubby account with two-step verification, notes, the Help menu, chat tabs and code search.

Helm: edit values, dry-run, upgrade, rollback and install from a chart.

Ask Kubby: ask questions about your cluster.

Preferences: appearance, tables, terminal, clusters and data.

Integrations: Argo CD, Helm releases and charts, pod and node metric charts.

Groups and drag-and-drop ordering in the cluster bar.

A Lens-style resource navigator.

Safe editing, scaling and other actions; PROD protection and audit log.

Command palette and terminal profiles.

Pod logs, describe, exec, port forwarding and the cluster terminal.

Overview, Nodes, Pods and Events: problems first.

Kubernetes engine and resource browser.

Cluster management: kubeconfig import and connection.

App shell and design system; dark and light themes.