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

Configure Cursor AI code editor to use agentgateway running in Kubernetes

Configure Cursor, the AI code editor, to route requests to your LLM through your agentgateway proxy.

Before you begin

1. Set up an agentgateway proxy.
2. Set up access to the OpenAI LLM provider.

Get the gateway URL

export INGRESS_GW_ADDRESS=$(kubectl get svc -n agentgateway-system agentgateway-proxy \
  -o jsonpath="{.status.loadBalancer.ingress[0]['hostname','ip']}")

echo "Gateway address: $INGRESS_GW_ADDRESS"

kubectl port-forward -n agentgateway-system svc/agentgateway-proxy 8080:80

Configure Cursor

1. Open the Cursor Settings.

   • macOS: Cmd + , or Cursor → Settings
   • Windows/Linux: Ctrl + , or File → Preferences → Settings

2. Navigate to the Models tab.

3. Enable Override OpenAI Base URL and enter your gateway address with the /v1 path. Cursor adds /chat/completions to it.

   http://<INGRESS_GW_ADDRESS>/v1

   For example, for a gateway on the ai-example.com secured host, use https://ai-example.com/v1. This base URL needs the route from the OpenAI v1/chat/completions tab of the OpenAI provider guide, which forwards every path to the backend. The Custom route tab rewrites every request to /v1/chat/completions, so it does not work with clients that add their own path, such as /chat/completions.

   Note

   You do not need to provide LLM provider credentials (such as an API key) through Cursor. The credentials are configured in agentgateway. Toggle off any API key overrides in Cursor.

Verify the connection

1. Open the Cursor chat panel (Cmd + L on macOS, Ctrl + L on Windows/Linux).
2. Send a message such as “test”.
3. Cursor responds through your agentgateway backend.