Get it for macOS

Shows your Kubernetes clusters in the bb sidebar, so you can spot a failing pod, read its logs and restart it without leaving the thread. Agents that deploy post a live card in the thread that shows which deployments restarted and which pods are still old. ArgoCD sync and health, node CPU, memory and disk, and PVC fill levels are one click away, along with port-forwards to any pod or service. Runs your own kubectl and kubeconfig; live CPU and memory need metrics-server in the cluster.


Overview

View source

What you get

  • A Kubernetes entry in the bb sidebar. Its badge counts failing pods and nodes that are not ready, so you notice trouble without opening anything.
  • A sidebar toolbar while the panel is open: kubeconfig context, a multi-select namespace filter with pod and problem counts per namespace, one name search shared by every page, cluster health, and the port-forwards you have running.
  • Pods with Logs and Restart buttons on every row, a jump to the owning Deployment, live CPU and memory, sorting, and a problems-only filter.
  • Deployments, StatefulSets and DaemonSets with their pods, logs from every pod merged by time, rollout restart, scaling and port-forward.
  • ArgoCD: sync and health of every Application, the resources that drift, Sync (prune optional), Refresh and Hard refresh, and a link to the ArgoCD web UI. It reads the Application objects with kubectl, so no ArgoCD token or CLI is needed.
  • Nodes: real CPU, memory and disk usage against requests and allocatable, node pools, memory overcommit, the PVCs mounted on each node, and the pods using most CPU and local disk.
  • Storage: how full every PVC is, which pods mount it, totals per storage class, claims nothing mounts, and released volumes still costing money.
  • Port-forwards to any pod, Deployment or Service on 127.0.0.1, with open and copy buttons.
  • Rollout cards in chat: when an agent deploys, it writes a kube-rollout block and the thread shows a live card per workload: restarted, still restarting, not restarted yet, or failed, with the new and old pods. The panel's Rollout check builds the same card by hand.
  • Jobs and CronJobs, Services and Ingresses, warning events, recently updated workloads, and kubectl describe for every object.
  • English and Russian interface, switchable from the toolbar.

For agents

The bb kube command gives agents the same view: status, pods --unhealthy, workloads, events, nodes, pvc, argocd and rollout, each with --json. A bundled skill tells agents to add a rollout card whenever they deploy.

How it works

The plugin runs your own kubectl with your kubeconfig, so any authentication your kubeconfig already uses keeps working. It always passes an explicit context and never changes your current context. Nothing leaves your machine except the kubectl calls to your clusters.

Every action that changes a cluster asks first and names the context. Turn on read-only mode in the plugin settings to hide restarts, scaling, job runs and ArgoCD syncs.

Requirements

  • kubectl on the machine running bb, plus any credential plugin your kubeconfig uses.
  • metrics-server in the cluster for live CPU and memory. Requests and limits work without it.
  • Read access to nodes/proxy for disk and PVC fill levels.
  • ArgoCD is optional and detected automatically.

More in Utilities 35

Inspect or control the computers bb runs on.

View all