OpenClaw Operator

The OpenClaw Operator runs OpenClaw AI agents on Kubernetes with production-grade security, observability, and lifecycle management.

A single OpenClawInstance custom resource defines the whole stack: StatefulSet, Service, RBAC, NetworkPolicy, PVC, PodDisruptionBudget, Ingress, and ServiceMonitor. The operator reconciles that resource into a hardened, monitored, self-healing deployment.

For the project pitch and a feature overview, see the README.

Where to start

• Deployment: install the operator on Kind, GKE, EKS, AKS, or any conformant Kubernetes cluster.
• Architecture: how the operator reconciles OpenClawInstance resources.
• Full example: an end-to-end working configuration.
• Custom providers: wire any AI provider via environment variables.
• API reference: every field on every CRD.

Operations

• Backup and restore
• Troubleshooting
• Runbooks for paging-worthy alerts live under runbooks/.
• Prometheus alert rules and Grafana dashboards live under monitoring/.

Self-configuration and adaptation

• Cluster defaults
• Self-config workflow
• Model fallback
• External secrets