stormglass
Download

Features

What Stormglass 0.2.1 does today, grouped by area. Items marked Coming soon are not in the app yet.

  • Every context in your kubeconfigSwitch from the sidebar; the last one is remembered.
  • Exec auth pluginsEKS, GKE and AKS credentials work as they do in your shell.
  • Cluster overviewNodes, namespaces, CRDs, and requests against allocatable CPU, memory and pods.
  • Workloads overviewCounts and health per kind, recent warnings, recent restarts.
  • Abnormal usageContainers near their limits, when metrics-server is installed.
  • Kubernetes version in the status barAlways visible, with the current context.
Cluster overview with node, namespace and CRD counts and a resource usage card showing pod, CPU and memory requests
  • Live listsWatches stream adds, changes and deletes. No polling interval.
  • Multi-namespace selectOne namespace, any set of them, or all.
  • Sort, filter, columnsPer resource kind.
  • Linked namesOwners, selectors, volumes and node names open their objects.
  • ReferencesEach object lists the ConfigMaps, Secrets, accounts and claims it uses.
  • EventsOn every detail page, and a cluster-wide list with warnings first.
  • Labels and annotationsCollapsible on every object.
  • Quick open⌘K jumps to any page or loaded object.
  • All built-in kindsWorkloads, networking, configuration, storage, quotas, policies and leases.
  • FavoritesStar objects to find them again.
  • Log viewerStreaming, with level detection, level picker and text filter.
  • Time range and downloadChoose how far back to read; save the log to a file.
  • TerminalA full terminal emulator for exec into any container.
  • FilesBrowse the container file system and download files.
  • Port forwardFrom the pod or service menu, to localhost.
  • Container infoImage, command, probes, ports, env and mounts.
  • Restart detailsRestart count, last restart time and reason on the list.
  • UsageCPU and memory against requests and limits, with metrics-server.
Pod log viewer with level picker, filter, time range and download buttons
  • YAML editorHighlighting, line numbers, cluster-owned fields dimmed and left out.
  • Server dry runAdmission and validation errors show before the real apply.
  • Create from templateStart a new object of any kind from a sensible skeleton.
  • ScaleDeployments, StatefulSets and ReplicaSets, up or down.
  • RestartRolling restart for Deployments, StatefulSets and DaemonSets.
  • Cordon and drainOn nodes, with a plain-language warning first.
  • CronJobsSuspend, resume, or trigger a run now.
  • Delete with contextThe confirmation says what else goes, such as a namespace's contents.
  • Helm releasesName, namespace, revision, status and last update.
  • CRDs in the sidebarGrouped by API domain, with the project's logo.
  • Custom resourcesList, detail, YAML and events like built-in kinds.
  • Logos stay privateBundled, or fetched from the project's own domain. Internal domains are never fetched.
  • Argo CD applicationsCounted on the cluster overview when installed.
  • Secret maskingValues are blurred in Secret details and container env.
  • Reveal and copyPer value, or copy all.
  • Confirm before writesEvery destructive action asks first.
  • Local onlyNo telemetry, no account, no relay server.
Secret detail with two data values blurred, each with reveal and copy buttons, and a Copy All button
  • ContainersState, CPU, memory, ports and status. Kubernetes containers hidden by default.
  • Images, volumes, networksListed with sizes and counts.
  • LogsThe same viewer as pods.
  • ShellTerminal into a running container.
  • FilesBrowse and download.
  • StatsLive CPU, memory, network and disk charts, and process count.
  • AWS Available
  • Google Cloud Coming soon
  • Azure Coming soon
  • DigitalOcean Coming soon
  • Spend overviewYesterday, month to date, forecast, and daily spend by service.
  • Top moversServices whose spend changed most this week.
  • Waste findingsIdle instances and databases, unattached volumes, old snapshots, unused IPs, idle load balancers and NAT gateways.
  • RightsizingSmaller instance types based on p95 usage.
  • AnomaliesDays above the expected range, with the likely cause.
  • BudgetsMonthly budgets with forecast against each.
  • Savings Plans and RIsCoverage and utilisation.
  • Kubernetes cost allocationBy namespace, workload and node, with idle requested capacity.
  • Approval-gated fixesShows the exact change, runs a dry run, then waits for Apply.
  • Action historyEvery applied fix is logged.
Cost recommendations with confidence, the metric behind each finding, and Apply, Ignore and Explain buttons (demo data)
  • Bring your own modelOllama, LM Studio, llama.cpp and vLLM found on localhost automatically.
  • Hosted modelsAny OpenAI-compatible API, or Anthropic. Keys stored in the system keychain.
  • Models in your clusterKServe inference services are discovered too.
  • Drives the UIOpens lists, filters and objects to show what it found.
  • Rich answersObject names in answers are links into the app.
  • Secret maskingSecret values are removed before anything is sent.
  • Approval for writesRead tools run freely; write tools wait for you. Per-tool permissions.
  • Session historyPast conversations and their tool calls are kept locally.
  • MCP serverFor Claude Code, Cursor and other clients. 127.0.0.1 only, token required.
  • Cost explanationsAsk why a finding was raised, from the finding itself.
  • Native RustGPU-drawn UI. No Electron, no web view.
  • macOS, Linux, WindowsOne codebase, same features.
  • ThemesCharcoal and Midnight dark, Paper and Silver light, six accents, or follow the system.
  • FontsPick the interface and monospace fonts, and scale the whole UI.
  • KeyboardBack and forward, quick open, and shortcuts for common actions.
  • OfflineWorks on air-gapped networks once licensed.

Missing something you need?

Tell us what you use every day that is not here yet.

Request a featureDownload