Headlamp 0.45.0 reduces desktop startup memory and unnecessary background requests, while adding new scheduling and Gateway API views, guided resource creation forms, and more flexible plugin and product customization. Security updates protect desktop APIs, verify external plugin archives, and close command-consent and dependency vulnerabilities. Accessibility improvements make high-zoom layouts, keyboard controls, and custom theme contrast more usable, alongside 31 bug fixes across authentication, port forwarding, Resource Map, tables, and the desktop app.

⚡ Performance

Plugin i18n now fetches only the active locale's translation file instead of all declared locales, preventing a flood of 404 requests when a plugin declares multiple languages.	Thanks to: @YousufFFFF @r0hansaxena. Thanks to @mjeanrichard for reporting. #7363
Desktop startup now defers optional work, retains smaller Kubernetes objects, and tunes garbage collection, reducing memory by about 60 MiB RSS across combined isolated measurements.	Thanks to @illume. #7358
Replaced the manual setInterval in useClustersVersion with React Query's useQueries, so polling pauses when the browser tab is hidden and unnecessary network requests to connected clusters are avoided.	Thanks to: @shreyas-acharya @illume. #6707

✨ Enhancements (part 1)

Added target container selection for ephemeral debug containers, allowing users to share the process namespace and filesystem of a specific container in multi-container pods during debugging.	Thanks to: @Suyog241005 @yush-1018. Thanks to @eightfourseventwo for reporting. #7245
Added a 'Keep Large Graph Groups Expanded' setting that persists namespace group expansion state across sessions, eliminating repetitive manual expansion clicks on large cluster resource maps.	Thanks to: @Suyog241005 @mudit06mah. Thanks to @syorito-hatsuki for reporting. #7156
A right-click context menu with copy and paste options has been added to the pod terminal, so users can copy and paste text without workarounds.	Thanks to @rootp1. Thanks to @jimmyjones2 for reporting. #7121
Added a form-based UI for creating PersistentVolumeClaims, including multi-select access mode support, so users can provision storage directly from the Storage section without writing raw YAML.	Thanks to @anoopw3bdev. #5333
Validated legal documents from the build manifest now appear in About and Legal tabs, letting desktop products present licenses and third-party notices in the app.	Thanks to: @illume @joaquimrocha. #7299
Exposed the host operating-system platform through the context-isolated desktop preload API, allowing frontend plugin authors to write platform-aware behavior in Electron builds.	Thanks to: @illume @sniok. #7291
Products can now configure custom titles and graphics for error and not-found pages at build time, allowing branded distributions to replace Headlamp's default error content without modifying source code.	Thanks to: @illume @joaquimrocha. #7288
The backend app name is configurable through --app-name or HEADLAMP_CONFIG_APP_NAME, letting branded distributions customize version output and User-Agent headers without maintaining a fork.	Thanks to: @illume @sniok. #7248
The Azure AKS plugin now receives its required script permissions from app configuration, allowing it to run without manual permission setup.	Thanks to @sniok. #7144

Target a specific container when debugging a pod

Choose which application container an ephemeral debug container targets, making its processes and filesystem available for more effective troubleshooting.

Cordoned and Drained node status indicators

See at a glance when a node is unavailable for scheduling instead of mistaking every Ready node for an active one.

TCPRoute and UDPRoute resource views

Browse and inspect Gateway API TCPRoute and UDPRoute resources directly alongside Headlamp's other Kubernetes resources.

Delete or evict pods

Choose eviction when maintenance should respect disruption safeguards, while retaining direct deletion when it is explicitly needed.

PodGroup and Workload scheduling views

Inspect scheduling resources and their status in dedicated views, making batch workload placement easier to understand.

✨ Enhancements (part 2)

Node list and detail views now display Cordoned and Drained status indicators, making it immediately clear when nodes are unschedulable without requiring users to inspect node conditions manually.	Thanks to @tejhan. Thanks to @sarg3nt for reporting. #7100
Plugins can replace or wrap the no-clusters Home page through registerClusterEmptyState, enabling custom onboarding while retaining Headlamp's default actions.	Thanks to: @illume @sniok. #7075
Pods without a scheduled node are now grouped under an Unscheduled node in the Resource Map when grouping by node, so pending pods remain visible instead of disappearing from the view.	Thanks to: @illume @gambtho. #7069
Desktop plugin and kubeconfig directories are now isolated by product name using a new --kubeconfig-dir option, preventing data collisions between different branded distributions sharing the same machine.	Thanks to: @illume @yolossn. #7068
JobSetList now probes the cluster for the JobSet API before rendering, showing a clear unavailable state with install docs when JobSet is not present instead of a confusing generic error.	Thanks to @Pritz395. #7016
TCPRoute and UDPRoute resources now have dedicated list and detail views in the Gateway API section, giving users full L4 route visibility alongside the existing HTTPRoute and GRPCRoute support.	Thanks to @areycruzer. #7002
GraphView and KubeIcon are now exported through pluginLib, allowing plugin authors to embed the resource map's graph renderer and icon set directly in their own pages.	Thanks to @rootp1. Thanks to @danbruno101 for reporting. #6992
The Service details view now lists the pods matched by the service selector, making it immediately obvious when a selector is misconfigured or targets no pods without manually cross-referencing endpoints.	Thanks to @krsatyamthakur-droid. Thanks to @jimmyjones2 for reporting. #6972
Hover-to-copy buttons have been added to IP and address table cells, allowing users to grab Pod IP, Cluster IP, External IP, and Endpoint addresses without manually selecting text.	Thanks to: @rootp1 @pallava-joshi. Thanks to @joaquimrocha for reporting. #7097

Terminal context menu

Copy and paste from a pod terminal using a familiar right-click menu instead of relying only on keyboard shortcuts.

Custom cluster empty states

Plugins can replace or extend the no-clusters screen, allowing distributions to provide onboarding that fits their environment.

Resource Map plugin APIs

Plugins can reuse Headlamp's graph view and Kubernetes icons to build visualizations that remain consistent with the rest of the interface.

High-resolution system tray icon

The desktop tray icon now stays sharp on high-density displays instead of appearing soft or pixelated.

StatefulSet creation form

Create StatefulSets through a structured form with guided fields instead of writing the complete manifest by hand.

<img width="1900" height="994" alt="frontend: CreateStatefulSetForm: Add create fo...

Headlamp is an easy-to-use and extensible Kubernetes web UI.

Need for speed! 🏎️ 💨  Headlamp 0.44.0 extends practical cluster support beyond 30,000 pods. Resource Map remains usable at this scale, Pod lists fetch 1,000 items at a time, Cluster Overview avoids unbounded watch traffic, and Brotli pre-compression reduces typical JavaScript and CSS transfers by 75-80% and server load for compression to 0%. Got dozens of clusters? Headlamp doesn't try to connect to all of them at startup anymore (potentially opening several browser login windows). Four new structured forms cover DaemonSets, CronJobs, Jobs, and ReplicaSets, while ConfigMap binaryData, ServiceAccount permission visibility, Gateway API v1.5.1, and new plugin controls expand day-to-day capabilities. The release includes 99+ bug fixes, including protections against white-screen startup failures when developing Headlamp on Windows, more accurate workload health reporting, stable large-manifest editing on large busy clusters, and actionable cluster connection errors. Accessibility work restores macOS zoom, improves WCAG AA contrast in a few spots, and broadens automated a11y checks. Security and documentation updates add github cosign of release checksums, safer redirects, tested in-cluster kubeconfig examples, and ready-to-use MCP configurations.

These changes were contributed by over 100+ people. Thanks! 🎉

⚡ Performance

ResourceMap performance improved with iterative BFS, graph simplification, and incremental WebSocket updates.

20,000+ pods — all operations under 37ms, where previously the browser would crash:

Added pagination to the Pods list with a 1,000-item budget and a load-more button, preventing out-of-memory crashes on clusters with tens of thousands of pods.

📱 Mobile

A mobile audit was done, and a number of improvements were made.

Tables now hide least-important columns first on narrow viewports instead of squashing them.

Mobile overflow menus and terminal dialogs now use the available screen space more effectively.

✨ Enhancements (part 1 of 2)

Added support for the binaryData field in ConfigMap detail and list views, so binary entries are now displayed, editable, and counted alongside data.

Added a structured creation form for DaemonSets, letting users create DaemonSet resources through the UI form alongside the existing YAML editor.

YAML syntax errors in EditorDialog now display a red squiggly underline at the exact line and column, making it easier to spot and fix invalid YAML before applying.

✨ Enhancements (part 2 of 2)

Headlamp is an easy-to-use and extensible Kubernetes web UI.

Headlamp is an easy-to-use and extensible Kubernetes web UI.

✨ Enhancements

• Added opt-in service account token auth for in-cluster deployments, enabling Headlamp to work behind OIDC or other external auth proxies. Thanks to @0xMH and @unixpariah. Also thanks to @yolossn for reporting the issue.
• Helm chart probes now support a configurable scheme (HTTP/HTTPS) and full timing settings, enabling correct probe behavior when backend TLS is enabled. Thanks to @gambtho. Also thanks to @mbasha86 for reporting the issue.
• Added alpha support for ClusterProfile discovery via the Cluster Inventory API, enabling automatic multi-cluster registration without manual kubeconfig setup. Thanks to @kahirokunn.
• Added hostAliases support to the Helm chart pod template, allowing custom host-to-IP mappings to be configured via chart values. Thanks to @VijayabaskarR-06. Also thanks to @geochip for reporting the issue.
• Added Russian language support to the frontend, making Headlamp more accessible for Russian-speaking users. Thanks to @Future998.
• Added proxy auth support so Headlamp can read authentication headers injected by middleware proxies for seamless cluster access. Thanks to @ageekymonk.
• Added dry-run preview for rollbacks, letting users inspect the resulting resource YAML before applying a rollout undo. Thanks to @alokdangre. Also thanks to @illume for reporting the issue.
• Added a node upgrade visualization panel showing Cordon, Drain, Delete, Reimage, Complete progress for managed nodes. Support for visualising upgrades in more k8s distros and systems coming soon. Thanks to @Jenniferyingni
• Added RTL layout support and new locales for Arabic, Urdu, and Hebrew, with the UI direction switching automatically based on the selected language. Thanks to @mahmoodalisha.
• Completed missing Hindi translations for UI strings, ensuring users with a Hindi locale see a fully translated interface. Thanks to @mahmoodalisha.
• Added a batch scale action to the multi-select toolbar, letting users scale multiple Deployments, StatefulSets, or ReplicaSets at once. Thanks to @RajPrakash681.
• Added registerProjectApiResource() to the plugin API, enabling plugins to include custom CRDs in Project resource fetching. Thanks to @NAME-ASHWANIYADAV.
• Added a diagnostics section to Pod and workload detail pages that surfaces deterministic troubleshooting hints from Kubernetes status, events, and container data. Thanks to @Utkarshpandey0001.
• Added a Dry Run option to the YAML editor, letting users validate manifests against the Kubernetes API before applying them. Thanks to @Vikaspal8923.
• Made the Back button on the cluster authentication chooser keyboard-accessible by adding tab focus and Enter/Space key handling. Thanks to @YadavAkhileshh.
• Added missing French translations, improving UI localization coverage and keeping Kubernetes terms like "cluster" technically accurate. Thanks to @codeurluce.
• Integrated @typescript/native-preview for frontend type checking, cutting type-check time from ~18-20 seconds down to ~3-4 seconds. Thanks to @sniok.
• Advanced Search queries can now be saved to browser local storage and restored, renamed, or deleted from the toolbar. Thanks to @Utkarshpandey0001.
• Added a dedicated Job details view surfacing completions, parallelism, backoff limit, active/succeeded/failed counts, and more, replacing the generic workload view. Thanks to @beep-boopp.
• Added remark-gfm to the headlamp-plugin do-not-copy list to prevent it from being incorrectly bundled into plugin builds. Thanks to @illume.
• Added ReadHeaderTimeout and IdleTimeout to the backend HTTP server, reducing slow-request attack risk while keeping logs and WebSockets unaffected. Thanks to @ayushmaan-16.
• Helm chart now supports setting a default pod debug image via values, used as the fallback when no per-cluster override is configured. Thanks to @harrshita123.
• Events table on the cluster overview page now includes a Node column showing which node emitted each event. Thanks to @VijayabaskarR-06. Also thanks to @trashhead for reporting the issue.
• Job details view now includes a Logs button, letting you view logs from related pods without leaving the Job page. Thanks to @ChunyiLyu.
• Added a container query parameter to pod log and exec deep-links, letting users link directly to a specific container's logs or terminal. Thanks to @jordanly.
• Resource detail views now show additional fields matching kubectl describe output, including port info for NodePort services. Thanks to @joaquimrocha.
• Added a Create Deployment form to the UI, letting users create Deployments directly alongside the existing Pod creation workflow. Thanks to @vyncent-t.
• Helm repository add and update operations now support authentication fields including username, password, and TLS options for private repositories. Thanks to @Athang69.
• Added --default-light-theme and related CLI flags so administrators can configure the default UI theme at runtime based on OS preference. Thanks to @guillaumebernard84.
• Added Windows Arm64 desktop builds, expanding native platform coverage to Arm-based Windows devices. Thanks to @illume.
• Added Bengali language support to the frontend and translated technical Kubernetes jargon, improving localization for Bengali-speaking users. Thanks to @Pratik-050.
• Node List and Details views now show the node pool, with a new getNodePool() method on the node model and Nodes added as a source to the Cluster group in the resource map. Thanks to @illume.
• Added an opt-in Chromium remote debugging port to the Electron app, enabling devtools and MCP integrations. Thanks to @yolossn
• Added a kubeconfig secretreader example to the Helm chart, pinned the Cluster Inventory provider images, and allowed Cluster Inventory commands to run under plugin mounts. Thanks to @kahirokunn.

Added alpha support for ClusterProfile discovery via the Cluster Inventory API

Added RTL layout support and new locales for Arabic, Urdu, and Hebrew

Added a node upgrade visualization panel

Added Russian language support to the frontend

Added proxy auth support for middleware proxies

Added a diagnostics section to Pod and workload detail pages

Job details view now includes a Logs button

Added a Create Deployment form to the UI

Added a dedicated Job details view

🐞 Bug Fixes

• Switched macOS notarization back to electron-builder, combining codesign, notarize, and staple into a single step using APPLEID credentials. Thanks to @joaquimrocha.
• Fixed OIDC login failures when no custom CA cert is configured, so public providers like Entra ID, Okta, and Google now work correctly. Thanks to @beep-boopp. Also thanks to @muhammad9970 for reporting the issue.
• Fixed parseRam and parseDiskSpace silently dropping decimal portions (e.g., '289.9Mi'), so memory and disk quantities are now parsed accurately. Thanks to @WasThatRudy. Also thanks to @thehandsomezebra for reporting the issue.
• Fixed the Electron app on Linux and Windows to fully quit when the window is closed instead of continuing to run in the background. Thanks to @Mohammed-Thaha. Also thanks to @Hronom for reporting the issue.
• Fixed stale Pod status icons in the list view that stayed yellow/warning even after a Pod had passed its readiness checks. Thanks to @RustamSheoran. Also thanks to @vikash7485 for reporting the issue.
• Fixed the Helm chart release workflow failing when it attempted to push provenance files as gzipped archives. Thanks to @onedr0p. Also thanks to @joryirving for reporting the issue.
• Table rows-per-page selection is now saved to localStorage, so the preference persists across page refreshes. Thanks to @ChayanDass. Also thanks to @illume for reporting the issue.
• Fixed missing edges in the resource map when nodes move between cloud and namespace-specific views. Thanks to @gattytto.
• Fixed a crash that prevented Headlamp from loading on Firefox browsers by removing problematic i18n normalisation. Thanks to @DhairyaMajmudar.
• Fixed a data race on the portforward Status and Error fields by introducing a mutex and thread-safe accessors, improving port-...

New release of the headlamp-plugin tool for plugin authors.

Upgrade

Upgrade your plugin to latest version by running:

npx @kinvolk/headlamp-plugin upgrade .

Bug fixes

• Fixed missing .cmd shim so the headlamp-plugin binary works correctly on Windows without requiring a shell wrapper.
• Replaced execSync with execFileSync in internal scripts to avoid shell injection risks.

Improvements

• headlamp-plugin create now scaffolds an AGENTS.md file into new plugins, and example plugins are bundled so AI coding agents have better context out of the box.
• Added missing type declarations to HeadlampTables, HeadlampChartStyles, and HeadlampHeaderStyle interfaces in additional.d.ts, improving type safety for plugins that customise themes.
• react-hooks ESLint rules are now enabled when linting plugins.

Dependency updates

• Security fixes: picomatch, vite, yaml, undici, tar, postcss, minimatch, qs, ajv, rollup, bn.js, terser-webpack-plugin, webpack, serialize-javascript, lodash, flatted.

Hitch hikers guide to Headlamp 0.42.0

Deep-link support was added for pod terminals (?view=exec) and log views (?view=logs), letting users bookmark or share direct links to specific UI states. Resources can now be searched by label in all list views, the Log Viewer gains a severity filter dropdown, and a delete button has been added directly to the plugins list. Node Details received two new sections for pod capacity usage and resource allocation, and the Resource Map now renders status badges on plugin-provided nodes and includes previously missing configuration resources. GRPCRoute details reach full parity with HTTPRoute, and a guided form-based resource creation UI (starting with Pods) is now available alongside the YAML editor.

On the reliability side, a silent cache desync that caused permanently stale cluster data was fixed, along with a lock-order inversion deadlock that could freeze WebSocket connections under load, and a wave of React hooks rule violations across more than a dozen frontend components. Several backend panics were fixed in port-forwarding, kubeconfig parsing, cache key generation, and Helm chart type handling, alongside debounced log aggregation to prevent UI freezes in All Pods mode. Security hardening includes path traversal prevention in the service proxy and dependency bumps.

✨ Enhancements (part 1)

Added a hidden 'labels' column to all resource list views so users can search resources by label (e.g. app=test) using the existing search bar.

Added a delete button directly to the plugins list page, so users can remove plugins without navigating to the detail page first.

Added a guided form-based UI for creating Kubernetes resources (starting with Pods), accessible from the sidebar and Pod list page alongside the existing YAML editor workflow.

✨ Enhancements (part 2)

Resource map graph nodes now support optional status metadata, allowing plugin-provided nodes to display warning and error badges like built-in resources.

Added ephemeral storage usage chart to the Node Details view, giving users visibility into ephemeral storage consumption alongside other node stats.

Added a download YAML button to the resource table, letting users export any resource manifest directly from the table view.

Improved global search UX by adding a clear button to the search input and showing a pointer cursor on hover over search results.

✨ Enhancements (part 3)

Added a pod capacity usage tile to the node details view so users can see how many pods are scheduled versus the node's allocatable pod limit.

Added a 'Review changes' tab to the resource editor dialog so users can diff the original manifest against their edits before applying.

Added UI support for the Kubernetes VolumeAttributesClass resource in the Storage section, including list and detail views.

Container port names are now displayed alongside protocol and port number in the containers list, providing more complete port information at a glance.

✨ Enhancements (part 4)

Extended a8r.io annotation support from Services to all Kubernetes resources, letting users see human-readable service discovery metadata on a...

Headlamp is an easy-to-use and extensible Kubernetes web UI.

Headlamp is an easy-to-use and extensible Kubernetes web UI.