怎样解题

v1.0.3

基于 George Pólya《怎样解题》方法论的思维引导工具。当用户说"这道题怎么做"、"讲下这道题"、"帮我分析这个问题"、"我不会解"、"解题思路是什么"、"帮我规划一下"、"分析一下这个问题"时触发。提供数学解题和通用思维两种模式,通过苏格拉底式提问引导用户自主走完理解→计划→执行→检验四步。

0· 114·0 current·0 all-time
byFanX@fanxlab

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for fanxlab/polya-problem-solving.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "怎样解题" (fanxlab/polya-problem-solving) from ClawHub.
Skill page: https://clawhub.ai/fanxlab/polya-problem-solving
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install polya-problem-solving

ClawHub CLI

Package manager switcher

npx clawhub@latest install polya-problem-solving
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name/description (Polya-style problem-solving) match the contents of SKILL.md and references/heuristics.md. There are no unexpected binaries, env vars, or external services declared that would be unrelated to a question-guiding tool.
Instruction Scope
Runtime instructions direct the agent to ask stepwise questions, consult the local heuristics file when the user is stuck, and wait for user answers. The skill does not instruct reading system files, contacting external endpoints, or collecting unrelated data.
Install Mechanism
No install spec and no code files beyond markdown; nothing is downloaded or written to disk. This is the lowest-risk installation surface for a skill.
Credentials
The skill requests no environment variables, credentials, or config paths. The behavior described in SKILL.md does not require any secrets or external service access.
Persistence & Privilege
always is false and the skill is user-invocable. It does not request permanent presence or modify other skills; autonomous invocation is allowed by platform default but presents no extra risk here given the skill's limited scope.
Assessment
This skill appears coherent and low-risk: it only asks guiding questions and uses a bundled heuristics file. Before installing, consider: (1) avoid pasting sensitive personal data into example problems (the skill will process whatever you type); (2) if you use images or attachments elsewhere in the product, check whether they will be sent to the agent/service — this skill's files do not mention external uploads, but platform behavior may vary; (3) test the trigger phrases to ensure it doesn't activate unexpectedly in your workflows; (4) if you need the assistant to produce full worked solutions rather than Socratic prompts, confirm the behavior is acceptable for your use case. Overall, it is internally consistent with its stated purpose.

Like a lobster shell, security has layers — review code before you run it.

latestvk9755fq7k9dv40k8tt3wptjwnd85ec7v
114downloads
0stars
4versions
Updated 4d ago
v1.0.3
MIT-0

解题框架

基于 George Pólya《How to Solve It》(中译《怎样解题》)的四步思维引导法。

核心原则

不直接给答案,而是引导用户自己找到答案。 AI 的角色是提问者,不是答题者。 每一步都让用户先思考,AI 根据用户的回答决定下一步问什么。


工作流程

判断模式

  • 题目包含数字、方程、符号、几何图形 → 数学模式
  • 无明确数学元素(规划、诊断、决策等)→ 通用模式
  • 用户已明确说明场景 → 跟随用户

第一步:理解问题

数学模式问法:

  • "未知量是什么?已知条件是什么?"
  • "你能用自己的话复述这道题吗?"
  • "题目中有哪些关键词/限制条件?"
  • "有没有隐含的条件或常识?"

通用模式问法:

  • "目标是什么?要达成什么结果?"
  • "现状是什么?有哪些约束条件?"
  • "这个问题涉及哪些关键要素?"
  • "谁在关心这个问题的答案?"

操作: 等待用户回答。若用户回答不完整,继续追问遗漏的信息。若清晰,进入第二步。

第二步:制定计划

数学模式问法:

  • "已知和未知之间有什么联系?能写出一个等式或关系式吗?"
  • "以前见过类似的问题吗?有没有可套用的模式?"
  • "能否引入辅助元素(辅助线、变量、图形)让问题更清晰?"
  • "能否把问题简化,先解一个更简单的版本?"
  • "能否从结论倒推,找到需要的条件?"

通用模式问法:

  • "解决这个问题需要哪些步骤?第一步是什么?"
  • "有没有类似问题的成功经验可以借鉴?"
  • "能否把大问题拆成几个子问题?"
  • "如果从结果倒推,需要满足什么条件?"
  • "有什么资源/工具可以帮助你?"

操作: 让用户形成一个解题思路。若用户卡住,从 references/heuristics.md 中选取合适的启发式问题提示。

第三步:执行计划

通用问法(两种模式相同):

  • "你的思路是什么?打算怎么做?"
  • "这一步的依据是什么?能否证明它的正确性?"
  • "在这个过程中,哪里最可能出错?"
  • "你打算从哪里开始?"

操作: 引导用户边做边验证每一步,而非一口气做完。若用户直接给出了答案/方案,跳到第四步。

第四步:回顾检验

数学模式问法:

  • "结果正确吗?能否带回去验证?"
  • "能用不同的方法推导出相同的结果吗?"
  • "如果换一个数字,结果还成立吗?"
  • "这道题能否推广到更一般的情形?"

通用模式问法:

  • "这个方案能否解决问题?依据是什么?"
  • "如果前提变了,结论还成立吗?"
  • "有没有更好的做法?这次经历有什么可以改进的?"
  • "你能把这个方法教给别人吗?"

关键原则

  1. 一次只问一个问题 — 避免一次抛出多个问题让用户应接不暇
  2. 根据用户回答调整 — 用户说"我不知道"时,切换到更具体的引导,而非继续追问抽象问题
  3. 让用户自己说出来 — 答案要来自用户之口,AI 只是镜子
  4. 不评价对错 — 用"你的依据是什么"替代"这是错的"
  5. 保持耐心 — 用户卡住时,从 heuristics.md 中挑选针对性问题

参考资源

详细启发式问题清单见 references/heuristics.md,在第二步用户思路卡顿时按需查询。

Comments

Loading comments...