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

Configure an agentgateway deployment as a model endpoint for kagent.

You can route model requests through an agentgateway deployment. Agentgateway is an AI-native proxy that provides traffic management, observability, and security for LLM calls. Because agentgateway exposes an OpenAI-compatible API, you configure the ModelConfig with provider: OpenAI and set openAI.baseUrl to your agentgateway Gateway service address.

Set up agentgateway model routing

1. Install agentgateway in your cluster. For more information, see the agentgateway documentation. Add --set agentgatewayModels.enabled=true to the Helm command for the agentgateway control plane.

2. Create a Gateway resource for model routing.

   kubectl apply -f - <<EOF
   apiVersion: gateway.networking.k8s.io/v1
   kind: Gateway
   metadata:
     name: agentgateway-proxy
     namespace: agentgateway-system
   spec:
     gatewayClassName: agentgateway
     listeners:
     - name: http
       protocol: HTTP
       port: 80
       allowedRoutes:
         namespaces:
           from: All
         kinds:
         - group: gateway.networking.k8s.io
           kind: HTTPRoute
         - group: agentgateway.dev
           kind: AgentgatewayModel
   EOF

3. Store the provider credentials that agentgateway uses to call the model. Agentgateway authenticates to the provider on your agents’ behalf, so this key belongs to the gateway rather than to kagent. The value is read from the Secret’s Authorization key by default.

   kubectl create secret generic openai-key -n agentgateway-system \
     --from-file=Authorization=<path-to-a-file-holding-your-api-key>

4. Create an AgentgatewayModel resource for each model that you want kagent to access. The resource name becomes the model name that kagent sends in requests, so it must match spec.model in the kagent ModelConfig. The following example routes requests for gpt-4o-mini to the OpenAI provider. For more provider options and authentication configuration, see the agentgateway model documentation.

   kubectl apply -f - <<EOF
   apiVersion: agentgateway.dev/v1alpha1
   kind: AgentgatewayModel
   metadata:
     name: gpt-4o-mini
     namespace: agentgateway-system
   spec:
     parentRefs:
     - group: gateway.networking.k8s.io
       kind: Gateway
       name: agentgateway-proxy
       sectionName: http
     provider: OpenAI
     policies:
       auth:
         secretRef:
           name: openai-key
   EOF

   Note

   policies.auth gives the gateway the credentials that it needs to reach the provider. A model with no auth is still accepted and programmed, and every request through it returns the provider’s own 401, because agentgateway forwards the call with no credentials.

5. Save the agentgateway Gateway service address in an environment variable. The /v1 suffix is required: kagent appends /chat/completions to this value, and agentgateway serves that endpoint at /v1/chat/completions. Without the suffix, every model call returns 404 Not Found.

   export AGENTGATEWAY_URL=http://agentgateway-proxy.agentgateway-system.svc.cluster.local/v1

Connect to an agentgateway endpoint

1. Create a ModelConfig resource. Choose the tab that matches your agentgateway authentication configuration.

   No authenticationAPI key authentication

   If your agentgateway deployment does not enforce any API key authentication, apply the following ModelConfig.

   kubectl apply -f - <<EOF
   apiVersion: kagent.dev/v1alpha2
   kind: ModelConfig
   metadata:
     name: agentgateway-model
     namespace: kagent
   spec:
     provider: OpenAI
     model: gpt-4o-mini
     openAI:
       baseUrl: "$AGENTGATEWAY_URL"
   EOF

   Review the following table to understand this configuration. For more information, see the API docs.

   Setting	Description
   provider	Set to OpenAI, because agentgateway exposes an OpenAI-compatible API.
   model	The model name to request from agentgateway. This value must match the name of an AgentgatewayModel resource in your agentgateway deployment.
   openAI.baseUrl	The Kubernetes service address of your agentgateway Gateway, including the /v1 path, set in $AGENTGATEWAY_URL.

   If your agentgateway deployment applies an apiKeyAuthentication policy with mode: Strict, you must provide an API key in the ModelConfig.

   1. Save the API key that the agentgateway apiKeyAuthentication policy requires. This is the key that clients must send in the Authorization header when they call agentgateway. Create a Kubernetes secret in the same namespace as your agent, such as kagent.

      export AGENTGATEWAY_API_KEY=<your-api-key>
      kubectl create secret generic kagent-agentgateway-key -n kagent \
        --from-literal AGENTGATEWAY_API_KEY=$AGENTGATEWAY_API_KEY

   2. Create the ModelConfig resource.

      kubectl apply -f - <<EOF
      apiVersion: kagent.dev/v1alpha2
      kind: ModelConfig
      metadata:
        name: agentgateway-model
        namespace: kagent
      spec:
        provider: OpenAI
        model: gpt-4o-mini
        apiKeySecret: kagent-agentgateway-key
        apiKeySecretKey: AGENTGATEWAY_API_KEY
        openAI:
          baseUrl: "$AGENTGATEWAY_URL"
      EOF

   Review the following table to understand this configuration. For more information, see the API docs.

   Setting	Description
   provider	Set to OpenAI, because agentgateway exposes an OpenAI-compatible API.
   model	The model name to request from agentgateway. This value must match the name of an AgentgatewayModel resource in your agentgateway deployment.
   apiKeySecret	The name of the Kubernetes secret that stores your agentgateway API key.
   apiKeySecretKey	The key in the secret that stores your API key.
   openAI.baseUrl	The Kubernetes service address of your agentgateway Gateway, including the /v1 path, set in $AGENTGATEWAY_URL.

2. Verify that the ModelConfig is accepted.

   kubectl get modelconfig agentgateway-model -n kagent -o yaml

Agentgateway is now added as a model endpoint in kagent. Next, you can create or update an agent to use this model.