智能回答框架 / Answer Framework
这个技能做什么 / What this skill does
- 帮助代理把开放问题整理成结论明确、证据充分、表达自然的回答。
Helps the agent turn open-ended questions into answers that are clear, evidence-based, and natural to read. - 适合解释、比较、观点表达、建议给出、方案分析等任务。
Works well for explanation, comparison, opinion, recommendation, and analytical discussion tasks. - 支持中文、英文,以及中英混合语境。
Supports Chinese, English, and mixed bilingual conversations.
什么时候使用 / When to use this skill
在以下情况优先使用:
Prefer this skill when:
- 用户在问“为什么”“如何”“你怎么看”“A 和 B 怎么选”。
The user asks “why,” “how,” “what do you think,” or “which option should I choose.” - 用户要的是解释、判断、分析、建议,而不是直接修改文件。
The user wants explanation, judgment, analysis, or recommendations rather than direct file edits. - 回答需要兼顾逻辑、可读性和可信度。
The answer should balance reasoning, readability, and credibility.
工作流程 / Workflow
- 先判断问题类型:事实型、解释型、对比型、观点型,或需要先澄清。
First identify the question type: factual, explanatory, comparison, opinion, or clarification-needed. - 尽早给出结论,不要把结论埋在最后。
State the conclusion early instead of burying it at the end. - 用 1 到 3 个关键证据或理由支撑结论。
Support the conclusion with 1 to 3 key pieces of evidence or reasons. - 明确写出“为什么这些证据会推出这个结论”。
Explicitly explain why those facts or reasons lead to the conclusion. - 根据用户语气控制长度和风格:简洁、详细、对比、观点。
Match the tone and depth to the user’s intent: concise, detailed, comparison, or opinionated. - 如果需求模糊,先提一个澄清问题,不要默默假设。
If the request is ambiguous, ask a clarifying question instead of silently choosing an interpretation.
回答模板 / Response patterns
事实型 / Factual
- 先给事实结论。
Start with the factual answer. - 再补一个关键来源、数据或上下文。
Then add one supporting detail, source, or context.
解释型 / Explanatory
- 先回答“是什么原因”。
First answer the “why” directly. - 再解释机制或因果链。
Then explain the mechanism or causal chain. - 必要时给一个例子。
Add an example if it improves understanding.
对比型 / Comparison
- 先说共同点或评判维度。
Start with common ground or comparison criteria. - 再列关键差异。
Then list the important differences. - 最后给出按场景的选择建议。
End with a recommendation based on specific needs.
观点型 / Opinion
- 明确立场。
State a clear position. - 给出两到三个论据。
Provide two or three arguments. - 补一个平衡视角或反方考虑。
Add a counterpoint or balancing consideration.
需要澄清 / Clarification-needed
- 只问 1 到 2 个最关键的问题。
Ask only 1 or 2 high-value clarification questions. - 不要一次丢给用户一串问卷。
Do not overwhelm the user with a long questionnaire.
输出要求 / Output expectations
- 回答必须直接回应问题,避免兜圈子。
The answer must directly address the question. - 证据要具体,避免空泛口号。
Evidence should be concrete, not vague. - 推理链要可读,不要只堆事实。
The reasoning chain should be visible, not just a dump of facts. - 尽量用自然过渡词,不要机械写成“答案:证据:推理:”。
Prefer natural transitions over rigid labels like “Answer / Evidence / Reasoning.” - 区分事实、推测、观点。
Distinguish facts, inferences, and opinions.
常见坑 / Common pitfalls
- 只给结论,不给依据。
Giving a conclusion without justification. - 只给很多信息,但没有清晰立场。
Providing lots of information without a clear takeaway. - 用户要简短回答,却写成长文。
Writing a long essay when the user asked for a short answer. - 问题含糊时擅自脑补。
Making silent assumptions when the request is underspecified.
简短示例 / Short example
用户 / User:Python 和 JavaScript 哪个更适合初学者?
回答思路 / Approach:
- 先给结论:如果重视语法直观和入门平滑,通常先学 Python。
Start with the conclusion: Python is usually easier for a smooth first step. - 证据:语法更简洁、样板更少、初学资料丰富。
Support it with evidence: simpler syntax, less boilerplate, lots of beginner material. - 补充条件:如果目标是前端网页开发,JavaScript 会更直接。
Add the condition: if the goal is front-end web development, JavaScript may be more direct.
记住:先回答,再论证;先清晰,再华丽。
Remember: answer first, justify second; clarity beats ornament.
Scan to join WeChat group