Documentation #26: Addition of MCP BE Service docs in deployment guides

}'

```

### 8. Configure oslmcpbackend (MCP Backend Service)

Change the respective fields:

- To configure the Ollama LLM server URL and model.

- To customize the AI assistant's system prompt and temperature.

- If you made changes to keycloak credentials.

- If you want to change logging level (TRACE / DEBUG / INFO / WARN / ERROR).

> ***If you are using a non-local domain, replace everywhere the http://keycloak:8080 with the respective {{protocol://domain.name}}.***

In folder `org.etsi.osl.main/compose/` edit the file `docker-compose.yaml`

```yaml  

  # AI Configuration

  SPRING_AI_OLLAMA_BASE_URL: http://ollama:11434

  SPRING_AI_OLLAMA_CHAT_MODEL: gpt-oss:20b

  SPRING_AI_OLLAMA_CHAT_TEMPERATURE: 0.5

  SPRING_AI_CHAT_SYSTEM_PROMPT: "You are an OpenSlice AI Assistant."

  SPRING_AI_CHAT_MAX_MESSAGES: 100

  # OAuth2/Keycloak Configuration

  SPRING_SECURITY_OAUTH2_RESOURCESERVER_JWT_ISSUER_URI: http://keycloak:8080/auth/realms/openslice

  SPRING_SECURITY_OAUTH2_CLIENT_PROVIDER_KEYCLOAK_ISSUER_URI: http://keycloak:8080/auth/realms/openslice

  SPRING_SECURITY_OAUTH2_CLIENT_REGISTRATION_KEYCLOAK_CLIENT_ID: osapiWebClientId

  # Logging Configuration

  LOGGING_LEVEL_ROOT: INFO

  LOGGING_LEVEL_OSL: INFO

  LOGGING_LEVEL_SPRING_AI: INFO

```

## Configure nginx

For this, you will have to replicate the steps in [Standalone CRIDGE deployment](#standalone-cridge-deployment) for every Cluster. Each CRIDGE instance will be in charged with the management of one Kubernetes Cluster.

### MCP Backend Service 

The MCP Backend Service provides AI-powered assistance using Ollama and connects to the OpenSlice MCP server.

To configure the MCP Backend Service, update the following fields in the `values.yaml` file:

```yaml

mcpbackend:

  enabled: true

  logLevelRoot: INFO

  logLevelOSL: INFO

  spring:

    logLevel: INFO

    ai:

      ollama:

        model: "gpt-oss:20b"           # Change the used model here

        temperature: 0.5

        apiUrl: "http://ollama:11434"  # Change the Ollama API URL here

      chat:

        systemPrompt: "You are an OpenSlice AI Assistant."  # Customize your initial Assistant prompt

        maxMessages: 100  # Maximum number of messages to keep in context

```

**Key Configuration Fields:**

- `mcpbackend.enabled`: Set to `true` to deploy the MCP Backend Service, `false` to disable it.

- `mcpbackend.spring.ai.ollama.apiUrl`: URL of your Ollama server. Update this if using an external Ollama instance.

- `mcpbackend.spring.ai.ollama.model`: The AI model to use (e.g., gpt-oss:20b, llama2, mistral).

- `mcpbackend.spring.ai.chat.systemPrompt`: Customize your initial Assistant prompt.

- `mcpbackend.spring.ai.ollama.temperature`: Controls randomness (0.0-1.0). Lower values are more deterministic.

- `mcpbackend.spring.ai.chat.maxMessages`: Maximum number of messages to keep in the conversation context.

> **Note:** Ensure the Ollama server is running and the specified model is available before starting the service.

### External Services (optional)