Preparing LiteLLM access

This section describes how to configure the IQ Integrator to authenticate with your LiteLLM proxy server using a static API key. This configuration allows the integration service to access your designated LLM providers and models.

Prerequisites

Deploy the IQ Integrator and MCP Server services.
If you require session persistence and multi-pod coordination, prepare the PostgreSQL database. If you use a static LiteLLM API key instead, the database is not required but is recommended to persist conversations and sessions.

Using a static LiteLLM API key

The IQ Integrator uses the static LiteLLM API key to authenticate outgoing API requests sent to your LiteLLM proxy server. This key allows the integration service to access the configured LLM providers and models through the proxy. To use this path, you must have the following components:

A deployed and running LiteLLM proxy server
Two configured proxy models (iq-general-purpose and iq-summary)
A LiteLLM API key from your proxy administrator

Complete the following steps to configure and verify the static key:

Create the LiteLLM API key secret in your namespace. The IQ Integrator reads the key from a Kubernetes secret named iq-litellm-api-secret:

kubectl create secret generic iq-litellm-api-secret \
  --from-literal=virtualKey="<YOUR_LITELLM_API_KEY>" \
  --namespace <DX_NAMESPACE>

Replace <YOUR_LITELLM_API_KEY> with the key provided by your LiteLLM proxy administrator.

If the secret already exists from a previous installation, delete the secret first and then recreate it:

kubectl delete secret iq-litellm-api-secret -n <DX_NAMESPACE>
kubectl create secret generic iq-litellm-api-secret \
  --from-literal=virtualKey="<YOUR_LITELLM_API_KEY>" \
  --namespace <DX_NAMESPACE>

Upgrade the IQ deployment:

helm upgrade dx-iq \
  https://<YOUR_REPOSITORY_FQDN_AND_PATH>/<IQ_HELM_CHART_VERSION>.tgz \
  --namespace <DX_NAMESPACE> \
  --reuse-values \
  --set configuration.litellm.liteLlmUrl="<YOUR_LITELLM_URL>"

Replace <YOUR_LITELLM_URL> with the URL of your LiteLLM proxy server. You can omit this flag if you already set the URL during the initial installation of the IQ Integrator.

Monitor the rollout status of the deployment:

kubectl rollout status deployment/dx-iq-integrator -n <DX_NAMESPACE>

Run the following command to check the logs for static key confirmation:

kubectl logs -n <DX_NAMESPACE> deployment/dx-iq-integrator | grep -i "LITELLM_API_KEY configured statically"

Confirm that the output matches the following log entry:
```
[DeploymentKey] LITELLM_API_KEY configured statically, skipping KMS flow
```
If you see this message, your static LiteLLM API key is correctly configured, and KMS flow was skipped.
Send a test message through the IQ chat interface to verify that you receive an AI-generated response without errors.

Related information