Hello, Could OpenAI provide clearer public guidance on task boundaries and refusal explanations in Codex and related agent experiences? The distinction I would like clarified is between searching for information, making recommendations, providing external…