AI Providers

AI providers enable the integration of AI agents into Sentinel for advanced data classification and analysis.

Currently supported features:

• AI Security Rules

If you do not plan to use the feature listed above, you can skip configuring an AI provider.

Supported Providers

Sentinel supports the following AI providers:

• Anthropic
• Azure OpenAI
• Google Generative AI
• Mistral
• Ollama
• OpenAI

Configuration

Configure your AI provider using the following environment variables:

• AI_PROVIDER: The name of the provider. Supported values: anthropic, azure, google, mistral, ollama, openai
• AI_PROVIDER_MODEL: The name of the model (e.g., gpt-4-turbo)
• AI_PROVIDER_OPTIONS: A JSON-encoded string of additional options, including your API key and optional settings like baseURL.
• AI_PROVIDER_REQUEST_OPTIONS: A JSON-encoded string of additional parameters to be sent in the API request (such as max_tokens, temperature etc.).

Example Configuration

AI_PROVIDER=mistral
AI_PROVIDER_MODEL=mistral-large-latest
AI_PROVIDER_OPTIONS={"apiKey":"abc123..."}

To specify a custom base URL:

AI_PROVIDER_OPTIONS={"apiKey":"abc123...","baseURL":"https://api.openai.com/v1"}

Provider-Specific Options

Common Options

• apiKey: API key for accessing the provider’s service.
• baseURL: (Optional) Custom API endpoint.

Azure OpenAI

• resourceName: The Azure resource name. Used to construct the base URL in the format: https://{resourceName}.openai.azure.com

OpenAI

• organization: (Optional) OpenAI organization ID.
• project: (Optional) OpenAI project ID.
• compatibility: Set to compatible if using a third-party provider that implements the OpenAI API.

Adjusting Request Parameters

To adjust request parameters such as max_tokens or temperature, configure the AI_PROVIDER_REQUEST_OPTIONS environment variable. Any parameters set in the request options will be added to (or override) parameters in the API request body.

Example:

AI_PROVIDER_REQUEST_OPTIONS={"max_tokens":300,"temperature":1,"stop":["\n\n"]}