Amazon Nova chat model integration.

Amazon Nova models are OpenAI-compatible and accessed via the OpenAI SDK pointed at Nova's endpoint.

Base exception for all Nova-related errors.

Exception raised for invalid request parameters (HTTP 400).

This includes:

• Invalid model parameters (temperature, top_p, etc.)
• Malformed requests
• Invalid tool definitions
• Invalid message form

Exception raised when model is not found or not available (HTTP 404).

This occurs when:

• Model name is invalid
• Model is not available in the region
• Model access has not been granted

Exception raised when rate limit is exceeded (HTTP 429).

This occurs when:

• Too many requests in a time window
• Token rate limit exceeded
• Request quota exceeded

Exception raised for internal model errors (HTTP 500).

This occurs when:

• Model encounters an internal error
• Service is temporarily unavailable
• Unexpected model behavior

Exception raised for tool calling errors.

This occurs when:

• Model doesn't support tool calling
• Tool definition is invalid
• Tool response format is incorrect

Exception raised for configuration errors.

This occurs when:

• API key is missing or invalid
• Base URL is invalid
• Environment variables are not set properly

Convert a tool to Nova's tool format.

Nova uses OpenAI-compatible tool format. This function handles conversion from LangChain tools, Pydantic models, or raw dicts.

Map HTTP/OpenAI errors to Nova-specific exceptions.

LangChain integration for Amazon Nova.

Amazon Nova chat models.