步骤 1：模糊的提示

initial_prompt = "Explain how to improve Python performance."
response = ask_claude(initial_prompt)
print("Initial Response:\n", response)

上述提示可能会产生一个模糊的答案，例如：

"要提高 Python 性能，你可以优化你的代码，使用高效的算法，并利用外部库。"

这个答案虽然正确，但缺乏可操作的细节。现在让我们澄清提示，让 Claude 提供更具体、更完整的回答。

refined_prompt = """
Explain five specific, verifiable methods to improve Python performance.

For each method, include:

- A short description

- A measurable example

- A small code snippet

Return your answer in a structured format.
"""
response = ask_claude(refined_prompt)
print("Clarified Response:\n", response)

通过添加数量词（五种方法）、输出结构和可衡量的示例，模型现在可以生成详细的、可验证的结果。

程序化自我澄清

在自动化流水线或聊天驱动的工作流中，可以指示 Claude 批判和澄清其自身的输出。这种自我检查模式大大提高了输出一致性和可解释性。

def clarify_response(original_prompt, model_output):
    """Ask Claude to review and clarify its previous response."""
    follow_up = f"""
The previous response may be incomplete or ambiguous.

Original prompt:

{original_prompt}

Response:

{model_output}

Task:

- Identify unclear or missing details.

- Provide a clearer, more complete version of the response.
"""
    return ask_claude(follow_up)