For the complete documentation index, see llms.txt. Markdown versions of all docs pages are available by appending .md to any docs URL.

Configure kagent to use models served through the SAP AI Core Orchestration Service.

kagent reaches SAP AI Core through its Orchestration Service, a single endpoint that serves models from several families, including Anthropic, OpenAI, Gemini, Amazon, Meta, and Mistral. Authentication uses OAuth2 client credentials from your SAP AI Core service key.

Create the ModelConfig

1. Save the OAuth2 client credentials from your SAP AI Core service key as environment variables.

   export SAP_AI_CORE_CLIENT_ID=<your_client_id>
   export SAP_AI_CORE_CLIENT_SECRET=<your_client_secret>

2. Create a Kubernetes Secret that stores both credentials. The Secret must contain the keys client_id and client_secret under exactly those names.

   kubectl create secret generic kagent-sapaicore -n kagent \
     --from-literal client_id=$SAP_AI_CORE_CLIENT_ID \
     --from-literal client_secret=$SAP_AI_CORE_CLIENT_SECRET

   Note

   SAP AI Core is the one provider that does not use apiKeySecretKey. kagent reads client_id and client_secret directly from the Secret that apiKeySecret names, so setting apiKeySecretKey has no effect.

3. Create a ModelConfig that references the Secret. The endpoint, resource group, and OAuth2 token endpoint all come from your SAP AI Core service key.

   kubectl apply -f - <<EOF
   apiVersion: kagent.dev/v1alpha3
   kind: ModelConfig
   metadata:
     name: sapaicore-model-config
     namespace: kagent
   spec:
     apiKeySecret: kagent-sapaicore
     model: anthropic--claude-4.5-sonnet
     provider: SAPAICore
     sapAICore:
       baseUrl: https://api.ai.prod.eu-central-1.aws.ml.hana.ondemand.com
       authUrl: https://<your-tenant>.authentication.eu10.hana.ondemand.com
       resourceGroup: default
   EOF

   Field	Description
   apiKeySecret	The name of the Kubernetes Secret that holds client_id and client_secret.
   model	The model to use, in the Orchestration Service naming convention, such as anthropic--claude-4.5-sonnet, gpt-5-mini, or gemini-3-pro-preview. For the full list, see the SAP AI Core models docs.
   provider	The provider to use, SAPAICore.
   sapAICore.baseUrl	The base URL for the SAP AI Core API. This field is required.
   sapAICore.authUrl	The OAuth2 token endpoint.
   sapAICore.resourceGroup	The resource group within SAP AI Core. Defaults to default.

   For every sapAICore field, including its type, default, and validation rules, see the API reference.

Use the ModelConfig

Reference the ModelConfig by name from an AgentTemplate in the same namespace.

spec:
  modelConfig:
    name: sapaicore-model-config

Next steps