What is KubeOpera
KubeOpera is an AIOps Kubernetes-as-a-Service (KaaS) platform that combines infrastructure management, intelligent observability, and autonomous AI agents. It is designed for platform engineering teams who need a system that can think, act, and learn — not just display dashboards.
What KubeOpera Does
KubeOpera combines intelligent workload management, real-time observability, automated remediation, and a full agentic AI layer that can reason, plan, and take action on your behalf. The rest of this documentation is organized around those pieces: Core Concepts for the architecture, Platform for the dashboard itself, Backend Services and Agentic AI for what's actually running underneath, and MCP Server for programmatic access.
Running the frontend locally
This is what most people mean by "getting started" — running the dashboard against an already-deployed backend, to develop against it or just try it out. Standing up a whole new KubeOpera environment from scratch is a different, infrastructure-level task: see the Setup Guide for how an environment is structured and how to provision one, either via GitOps or the standalone Helm chart.
git clone https://github.com/ochestra-tech/kubeopera
cd kubeopera
cp .env.example .env.local # fill in backend service URLs and credentials
npm install
npm run dev # starts Next.js on :3000
.env.example is worth reading in full rather than skimming — several variables have non-obvious behavior explained in its own inline comments, most notably the AUTH_APP_ID / NEXT_PUBLIC_AUTH_APP_ID pair: the former is read server-side on every request and takes effect immediately, while the latter is baked into the compiled bundle at build time and does nothing if changed afterward without a rebuild. Getting the two out of sync is a common source of confusing login behavior.
Accessing an already-deployed dashboard
If you're pointing at a real deployed environment rather than running locally, the frontend lives in the kubeopera-core namespace:
kubectl port-forward -n kubeopera-core svc/kubeopera-frontend 3000:80
# open http://localhost:3000
There's no environment-variable-driven admin bootstrap — the first platform administrator account is seeded directly into the database by a migration the first time auth-service starts against a fresh database, covered in the Setup Guide.
Service Port Reference
These are the real, currently-deployed container ports — useful when port-forwarding a specific service directly rather than going through the frontend's proxy.
| Service | Port | Purpose |
|---|---|---|
| kubeopera-frontend | 3000 | Next.js dashboard |
| kubeopera-api | 8090 | App, deployment, pipeline, and build management API |
| auth-service | 8081 | Identity, JWT issuance, RBAC |
| k8s-monitor | 8085 | Health, cost, metrics API |
| security-api | 8086 | Security posture API |
| cicd-gateway | 8087 | CI/CD webhook receiver, pipeline runs |
| anomaly-detector | 8088 | Anomaly events & self-healing |
| predictive-scaler | 8089 | Load forecasts & scaling |
| incident-manager | 8090 | Incident lifecycle |
| app-service | 8112 | Manifest generation preview |
| observability-agent-srv | 8092 | Telemetry collection agent |
| analysis-agent-srv | 8093 | Statistical analysis agent |
| action-agent-srv | 8094 | K8s action execution agent |
| feedback-agent-srv | 8095 | Outcome feedback agent |
| recommendation-agent-srv | 8096 | Claude-powered recommendations |
| agent-runtime | 8111 | Agent reasoning engine (SRE Orchestrator and others) |
| build-service | 8098 | Kaniko image builds from a Git repository |
| nodes-manager | 8115 | Node lifecycle, Karpenter integration |
| kubeopera-ai (Optimizer) | 8113 | Continuous health monitoring, on-demand optimisation |
| app-advisor-srv | 8105 | Per-app domain profile and advice |
kubeopera-api and incident-manager share port 8090 by coincidence of separate teams picking the same number early on — they're different services in different namespaces, so this has never caused a real conflict, but it's worth knowing if a port number alone doesn't uniquely identify a service in your notes.
Next Steps
- Core Concepts — understand the layered architecture
- Platform Guide — explore the dashboard and UI features
- Agentic AI — learn how the AI agents work
- MCP Server — connect Claude Desktop or Claude Code
- Setup Guide — standing up a new environment
Need Help?
- GitHub Issues: github.com/ochestra-tech/kubeopera/issues
- GitHub Discussions: github.com/ochestra-tech/kubeopera/discussions