Lumen / Releases
Lumen releases
Download the latest macOS installer and review what changed. Lumen 0.0.9 brings a live background resource cache, Crossplane views, deeper deployment detail and independent cluster terminals.
13 October 2026
Lumen 0.0.9 / Latest release
Release notes copied from the Lumen repository changelog, with UK English spelling. Dates follow the changelog.
Lumen 0.0.9
Highlights
- Background resource cache: deployments, pods, secrets, configmaps, services and Helm releases load once per cluster and stay live; tab and namespace switches no longer reload or restart watches.
- Crossplane section: read-only sidebar group (shown only when Crossplane is detected) with Overview, Packages, Definitions, Composites and Managed views, a detail drawer and a Claim → XR → MR graph.
- Deployment Full View: 12-section full-page deployment view with revision diff and security posture, opened from the drawer.
- Rename cluster contexts: right-click a cluster → Rename; kubeconfig, pins, tabs, watchers and terminals follow.
- Live kubeconfig refresh: contexts added to the kubeconfig appear without restarting Lumen.
- Multiple context-bound terminals: independent terminal tabs, each bound to the cluster context it was opened with.
Lumen 0.0.9
Resource cache
- Main-process LIST+WATCH streams per cluster with 650 ms coalesced batches into a normalised store.
- Cluster-wide watches with client-side namespace filtering; automatic per-namespace fallback on 403.
- Status banners for reconnecting, error and forbidden states while keeping rows visible.
- Resync on window focus and system wake.
- Helm releases derived from the secrets stream; release blobs never reach the renderer.
- Removed legacy per-event watch IPC and the old pod worker hook.
Lumen 0.0.9
Nodes
- New architecture distribution donut (ARM / non-ARM / unknown) that follows the search filter.
- Fixed the Nodes table being hidden or clipped with Stats expanded or in short windows.
Lumen 0.0.9
Bottom panel & terminals
- Unsaved YAML edits persist per tab across tab switches, pop-out and refreshes; discard confirmation on close.
- YAML tab IDs are cluster-scoped (fixes saving to the wrong cluster).
- Terminals bound to their opening context via a per-terminal override kubeconfig; context shown in tab subtitle, badge and banner.
- Multiple terminal tabs with a + action and Cmd+Shift+`.
- Cordon/drain use an explicit --context instead of switching the global context.
- PTYs are cleaned up on window close and quit.
Lumen 0.0.9
Fixes
- Preload event subscriptions now remove only their own listeners.
- Multi-namespace selection and stale generic watcher leaks fixed.
- Custom kubeconfig path in Settings is now honoured.
Lumen 0.0.9
Dependencies
- Electron 42.x (fix:pty targets 42.2.0).
- Added crossview Apache-2.0 licence notice.
29 April 2026
Lumen 0.0.8
Release notes copied from the Lumen repository changelog, with UK English spelling. Dates follow the changelog.
Lumen 0.0.8
Highlights
- Agentic AI debugging: the AI assistant can now execute kubectl commands against your cluster to gather real data for debugging, with human-in-the-loop approval.
- AWS settings split: AWS credentials and Granted integration moved to a dedicated settings section.
- Status bar cluster switcher: click the cluster name in the bottom bar to quickly switch clusters.
- Status bar AWS profile indicator: see and switch your active AWS profile from the bottom bar.
- Node metrics from metrics-server: NodesView now shows real CPU/memory usage instead of pod request estimates.
- Notification mute: mute anomaly toast notifications to reduce noise on high-churn clusters.
Lumen 0.0.8
Agentic AI (tool calling)
- AI assistant can execute kubectl commands (get, describe, logs, events, etc.) to debug issues.
- Human-in-the-loop approval: every tool call shows an Allow / Allow & Trust / Deny prompt.
- Trusted command prefixes auto-approve future calls (e.g. trusting kubectl get approves all get commands).
- Trusted commands visible and manageable in Settings → AI Models → Tool Calling.
- Safety: dangerous commands (delete namespace, exec, port-forward) are always blocked.
- Read-only and read-write modes configurable in settings.
- Works with Bedrock (Claude), Google Gemini, and local LLMs (LM Studio, Ollama).
Lumen 0.0.8
Settings & UI
- AWS credentials split out of AI settings into a dedicated AWS section with Granted detection.
- New AWS nav item in settings sidebar.
- Status bar: clickable cluster name opens a quick-switch popup.
- Status bar: AWS profile indicator with quick-switch popup, Granted-aware.
- Notification mute button in status bar: hides anomaly toasts, clears existing ones on mute.
- App logo fixed for production builds (Vite asset import) and macOS dock (rounded-corner alpha mask).
- Node delete no longer crashes the UI (optional chaining fix).
Lumen 0.0.8
Node monitoring
- NodesView CPU/Memory columns now show real usage from metrics-server (refreshed every 30s).
- Removed broken pod-request-based utilisation that always showed 0%.
- Column headers updated from CPU Requests / Memory Requests to CPU Usage / Memory Usage.
- Added getNodeMetrics IPC endpoint for metrics-server data.
Lumen 0.0.8
Dependencies
- Updated @ai-sdk/amazon-bedrock to 4.0.96, @ai-sdk/google to 3.0.64.
- Added @ai-sdk/openai-compatible for local LLM tool calling support.
- Added zod as direct dependency.
27 April 2026
Lumen 0.0.7
Release notes copied from the Lumen repository changelog, with UK English spelling. Dates follow the changelog.
Lumen 0.0.7
Highlights
- Revamped AI assistance.
- Account login support: sign in and registration for synced preferences and org-ready workflows.
- Local LLM support: OpenAI-compatible endpoints for private, on-device inference.
Earlier releases
The full GitHub release history includes earlier installers, dated notes and release-specific installation guidance.