Configuring Content Cortex AI Services components

Configure the reasoning service and MCP Server components to enable Content Cortex AI Services to process user requests and interact with content repositories.

About this task

The reasoning service orchestrates interactions between the AI Agent plug-in, MCP Server, and large language models (LLMs), while the MCP Server exposes content repository capabilities through the Model Context Protocol. Configuration for both components is managed through property files that you create and edit by using the prerequisites.py script, which then generates Kubernetes deployment artifacts.

For information about running the prerequisites.py script in gather, generate, and validate modes, see Generating simple custom resource and deployment files.

Procedure

  1. Use the prerequisites.py script in gather mode to create the property files for IBM Content Cortex AI Services.

    The script creates property files in the propertyFile directory.

  2. Edit the property files to configure the reasoning service and MCP server parameters.

    Edit the following property files to configure your deployment:

    • ccx-deployment.toml - General deployment settings for the reasoning service and MCP servers
    • ccx-identity_provider.toml - Identity provider (IDP) settings
    • aiservices_providers.toml - LLM provider settings and authentication
    • aiservices_integration.toml - Content deployment settings

    Complete all the required values in these files.

  3. Use the prerequisites.py script in generate mode to create the Kubernetes deployment files.

    The script generates the following artifacts in the generatedFolder directory:

    • ibm-ai-services-integration-config.yaml - ConfigMap containing IDP and content connection settings
    • ibm-providers-config-secret.yaml - Secret containing AI provider information
    • ibm_ai_services_cr_production.yaml - Custom resource that manages the deployment
    • ibm-idp-public-key-secret.yaml - Secret containing the IDP public key
    • Additional SSL certificate secrets that are based on your IDP and GraphQL endpoint configuration
  4. Use the prerequisites.py script in validate mode to verify and apply the generated files to your cluster.

    The validate mode checks database connectivity, LDAP connectivity, IDP configuration, and AI Services configuration before optionally applying the files.

    Alternatively, you can apply the generated files manually:

    kubectl apply -f ibm-idp-public-key-secret.yaml -n <namespace>
kubectl apply -f ibm-providers-config-secret.yaml -n <namespace>
kubectl apply -f ibm-ai-services-integration-config.yaml -n <namespace>
kubectl apply -f ibm_ai_services_cr_production.yaml -n <namespace>
    The operator uses these applied files to deploy and configure the Reasoning Service and MCP servers.
  5. Verify the reasoning service pod restarts with the new configuration. 
    kubectl get pods -n <namespace> -l app.kubernetes.io/name=ibm-reasoning-service
  6. Check the reasoning service logs to confirm the configuration is loaded correctly. 
    kubectl logs -n <namespace> <reasoning-service-pod-name>

    Look for configuration initialization messages that confirm whether the LLM provider, MCP server connections, and other settings are loaded.