openai-responses-api

To use the OpenAI Responses API in Python, call `client.responses.create` with the `input=` parameter instead of the legacy `client.chat.completions.create` method. This enforces correct request structures and leverages the updated API standards.

Format structured output with the OpenAI Responses API by using the `text={"format": ...}` parameter in your request. This guides the model to return data structured according to your specified format requirements.

Handle multi-turn conversations in the OpenAI Responses API by using the `previous_response_id` parameter. This maintains context across turns efficiently without manually passing the entire message history each time.

Manage model selection by using environment variables with safe defaults. If an API request returns a `NotFoundError`, the Skill suggests falling back to alternative available models to resolve the error.

Yes, you can use tool calling with the OpenAI Responses API. The API provides a specific tool-calling mechanism that allows the model to interact with external tools and functions as part of its response generation process.

Migrate to the OpenAI Responses API to improve compatibility and leverage new features like structured output formatting and efficient multi-turn conversations. It acts as a guardrail ensuring developers adhere to the latest API standards.