K8pilot is a production-grade, AI-powered Kubernetes dashboard built for DevOps engineers who demand more than kubectl. It combines real-time observability, security auditing, cost intelligence, and a proactive AI engine into a stunning, code-split, themeable interface with 54 views and 75+ API endpoints.
Docker Image:
cerebro46/k8pilot:latest
Traditional Kubernetes management is painful. We've all been there:
- Terminal Blindness: Staring at
kubectl get podsacross 20+ namespaces and missing the one pod that's quietly OOMKilled. - Troubleshooting Hell: The endless cycle of
logs→describe→google→repeat. - Reactive Ops: Waiting for a Slack alert after a pod has already been down for 5 minutes.
- Security Gaps: Deploying images with
:latesttags or privileged containers because "it just needs to work" for now. - Cost Blindsides: No visibility into resource waste until the monthly cloud bill arrives.
K8pilot turns this chaos into a clean, intelligent, and visually stunning control center — shifting from reactive monitoring to proactive intelligence.
| Feature | Description |
|---|---|
| Aura Hub | Unified application dashboard with namespace grouping and health overview |
| Cluster Pulse | Real-time CPU/Memory heatmap showing workload pressure across the cluster |
| Intelligence Feed | Central stream of scaling, restart, deletion, and network events |
| AI Pod Doctor | Deep root-cause analysis of pod failures with fix proposals |
| AI-Assisted Remediation | One-click YAML patches to resolve CrashLoops or OOMs |
| Namespace 360 | Namespace-level deep dive with resource topology |
| Feature | Description |
|---|---|
| Pod Health Matrix | Heuristic 0–100 health scoring for every pod with issue detection |
| Incident Timeline | Correlated K8s warning events with severity escalation and status tracking |
| Rollout Tracker | Real-time deployment rollout status with animated progress bars |
| CronJob Monitor | Scheduled workload tracking with execution history and success/fail rates |
| Resource Heatmap | Per-node CPU/memory intensity visualization |
| Network Diagnostics | DNS resolution audit, endpoint health, service connectivity mapping |
| Log Stream | Real-time pod log streaming with container selection |
| Feature | Description |
|---|---|
| Cluster Benchmark | Comprehensive A+ → D grading across Security, Reliability, Resources, Networking, Storage |
| RBAC Auditor | Cluster role and binding analysis with permission matrix |
| Image Scanner | Detects :latest tags, missing digests, public registry usage |
| Security Scorecard | Namespace-level GPA compliance grading |
| TLS Auditor | Certificate and secret audit with expiry tracking |
| Network Policies | Policy coverage and rule visualization |
| Feature | Description |
|---|---|
| Resource Recommender | Right-sizing analysis with estimated monthly savings |
| Capacity Planner | Cluster resource forecasting with optimization recommendations |
| Cloud Cost Profiler | Estimated burn rate by namespace based on live resource requests |
| Config Drift Detector | Cross-namespace ConfigMap/Secret comparison with per-key diffs |
| Cost Optimizer | Identifies over-provisioned and under-utilized workloads |
| Feature | Description |
|---|---|
| Nodes & Clusters | Node health, capacity, taints, conditions with detail panels |
| Node HA Spread | Pod distribution analysis across availability zones |
| Topology Map | Interactive cluster resource relationship graph |
| Aura Tree | Hierarchical namespace → deployment → pod drill-down |
| Ghost Inspector | Automated zombie resource detection (unused ConfigMaps, Secrets, PVCs) |
| Rollback Engine | Deployment revision history with one-click rollback |
| Network Listen | Live service-to-service traffic flow monitoring |
Every theme maintains the premium glassmorphism aesthetic:
| Midnight | Arctic | Catppuccin | Tokyo Night | Night Owl |
| Everforest | Rosé Pine | Ayu Mirage | Deep Sea | One Dark |
| Palenight | IBM Carbon | Monokai Pro | GitHub Dark | Shades of Purple |
| Vesper | Gruvbox | Dracula | Nord | Obsidian |
| Cobalt | Graphite | Forest | Abyss | Monochrome |
BEFORE (v3.5): 965 kB single bundle ⚠️ Over 500kB limit
AFTER (v4.1): 104 kB initial load ✅ 89% reduction
- 47 lazy-loaded view chunks — each view loads on-demand (2.8–47 kB each)
- 4 vendor bundles — React, Framer Motion, Lucide Icons, xterm cached independently
- Suspense fallback — skeleton loader while chunks load
- Categorized sidebar — 9 collapsible groups instead of 46 flat items
- Node.js/Express with 75+ REST API endpoints
- Shared utility modules —
parseCpu,parseMem,safeGetextracted intoserver/utils.js - TTL cache layer —
server/cache.jsfor K8s API response deduplication (10s TTL) - JWT authentication with 24h session tokens
- WebSocket streaming for logs and terminal
| Layer | Technology |
|---|---|
| Frontend | React 18, TypeScript, Vite |
| Styling | Vanilla CSS (Glassmorphism Design System) |
| Animations | Framer Motion |
| Icons | Lucide React |
| Backend | Node.js, Express |
| Kubernetes | @kubernetes/client-node (CoreV1, AppsV1, BatchV1, RBAC, Networking) |
| Terminal | xterm.js with WebSocket |
| Auth | JWT + bcrypt |
docker build -t cerebro46/k8pilot:latest .
docker push cerebro46/k8pilot:latestkubectl apply -f deploy/k8pilot-full.yamlImportant
The deployment manifest includes RBAC with full cluster-read permissions. Review deploy/k8pilot-full.yaml before applying to production clusters.
kubectl port-forward svc/k8pilot-service 5000:80 -n k8pilotOpen http://localhost:5000 — Default credentials: admin / admin123
# Terminal 1: Frontend (hot-reload)
npm install
npm run dev
# Terminal 2: Backend
cd server && npm install && node index.js| Variable | Description | Default |
|---|---|---|
JWT_SECRET |
Secret for session signing | k8pilot-super-secret-2026 |
ADMIN_USER |
Dashboard username | admin |
ADMIN_PASSWORD |
Dashboard password | admin123 |
K8pilot/
├── src/
│ ├── App.tsx # Main router with React.lazy() code-splitting
│ ├── index.css # Theme system (25 themes) + design tokens
│ ├── services/k8s.ts # Frontend API service layer
│ └── components/ # 54 view components (lazy-loaded)
├── server/
│ ├── index.js # Express API server (75+ endpoints)
│ ├── utils.js # Shared K8s parsing utilities
│ ├── cache.js # TTL cache with request deduplication
│ └── metrics.js # Metrics collector
├── deploy/
│ └── k8pilot-full.yaml # Full K8s deployment manifest
└── Dockerfile # Multi-stage build (Node 18 Alpine)
| Version | Codename | Highlights |
|---|---|---|
| v4.1 | Nova | Code-split architecture, 5 new features (CronJob Monitor, Config Drift, Resource Recommender, Network Diagnostics, Cluster Benchmark), categorized sidebar, TTL cache |
| v4.0 | Nova | Image Scanner, Rollout Tracker, Capacity Planner, Pod Health Matrix, Incident Timeline |
| v3.5 | Orion | Intelligence Feed, TLS Auditor, Ghost Inspector, Network Listen, 25 themes |
| v3.0 | Orion | AI Pod Doctor, Security Scorecard, RBAC Explorer, Cost Profiler |
Built with ❤️ by Priyesh. Feel free to open issues or submit PRs for new features or themes!