API reference

The Kubernetes API is the primary interface

Every workload, volume, route, policy and served model is a resource you can create, inspect and version. That means the canonical API reference for most of the platform is the Kubernetes API reference, plus the CRDs the platform installs.

Discover what is available on your own cluster:

# All resource types, including CRDs
oc api-resources

# The schema for a specific type
oc explain virtualmachine.spec --recursive

# The OpenAPI schema the cluster serves
oc get --raw /openapi/v2 > openapi.json

This is worth doing before reading any documentation: the cluster is authoritative about what it actually supports.

Authentication

Requests authenticate with a bearer token — either your user token or a service account token. See Access the platform for how to obtain each, and prefer short-lived service account tokens for anything automated.

curl -sS -H "Authorization: Bearer <token>" \
  "<your-cluster-api-endpoint>/api/v1/namespaces/<project>/pods"

Common resource groups

Group	Resources
core/v1	Pod, Service, Secret, ConfigMap, PersistentVolumeClaim
apps/v1	Deployment, StatefulSet, DaemonSet
networking.k8s.io/v1	NetworkPolicy, Ingress
gateway.networking.k8s.io	Gateway, HTTPRoute
kubevirt.io	VirtualMachine, VirtualMachineInstance
serving.kserve.io	InferenceService, ServingRuntime
k8s.ovn.org	UserDefinedNetwork

Inference APIs

Model endpoints and the LLM gateway speak an OpenAI-compatible API, so existing OpenAI SDKs work against them with a base URL change. See Deploy an LLM gateway.

Scaffold status

Platform-specific REST endpoints (billing, project provisioning, gateway admin) will be published from their OpenAPI specifications, rendered on this page rather than maintained by hand.