Zhang Xuefeng Perspective Skill: A Role-Play Persona for Education Advice
A role-play skill by Huashu that answers education and career questions in the style of Zhang Xuefeng, an imitation built from his public statements.
- Installs
- 11.5K
- on skills.sh
- Stars
- 10.7K
- on GitHub
- Updated
- Oct 2026
- last commit
The zhangxuefeng-perspective skill switches your agent into a role-play of Zhang Xuefeng, a Chinese education and exam-prep commentator. SKILL.md says he died on March 24, 2026, and that the persona is built from his public statements. Once active, the agent answers as him, in first person, in a fast, blunt, joke-driven style.
Before it answers about a specific major, school or industry, SKILL.md says the agent must research current facts with tools such as WebSearch. It then applies five mental models and eight decision heuristics, and usually asks about your family's situation and exam score first. It says once at first activation that this is an inference from public remarks, not his own opinion.
Good for
- Parents and students weighing college majors, schools and careers
- Anyone who wants a blunt, data-first second opinion on an education choice
- Readers studying how a public figure's reasoning could be written down as a skill
Install zhangxuefeng-perspective
Claude Code
- 1
Copy this prompt.
Install the zhangxuefeng-perspective skill in Claude Code. Run this command: npx skills add alchaincyf/zhangxuefeng-skill Then check that it loaded and tell me how to use it.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
Requires Node.js.
The README's manual route clones the repo into the skills folder. Only a git clone lets the skill's monthly update check run.
From the project README. Source.
What's in the skill
When it loads
张雪峰的思维框架与表达方式。基于5本著作、15+篇权威媒体深度采访、 30+条一手语录、11个关键决策记录和完整人生时间线的深度调研， 提炼5个核心心智模型、8条决策启发式和完整的表达DNA。 用途：作为思维顾问，用张雪峰的视角分析教育选择、职业规划、阶层流动等问题。 当用户提到「用张雪峰的视角」「张雪峰会怎么看」「张雪峰模式」「雪峰视角」时使用。 即使用户只是说「帮我用张雪峰的角度想想」「如果张雪峰会怎么说」「切换到张雪峰」也应触发。
Sections in SKILL.md
- 角色扮演规则（最重要）
- 回答工作流（Agentic Protocol）
- Step 1: 问题分类
- Step 2: 张雪峰式研究（按问题类型选择）
- Step 3: 张雪峰式回答
- 🔴 CHECKPOINT · 开口前三问
- 失败模式与 Fallback 树
- 示例：Agentic vs 非Agentic
- 身份卡
- 核心心智模型
- 模型1: 社会筛子论
- 模型2: 选择 > 努力
- 模型3: 就业倒推法
- 模型4: 阶层现实主义
- 模型5: 争议即传播
- 决策启发式
Key facts
- Installs on skills.sh
- 11,531 as of 2026-11-02
- License
- MIT
- Last repo commit
- 2026-10-13
What it includes
Answers in first person as Zhang Xuefeng until you ask it to stop.
A disclaimer is given once at first activation. You exit by saying phrases like 退出 or 不用扮演了.
Read ·
Role-play rulesClassifies the question, researches facts, then answers.
Fact questions about majors, schools or industries go to research first. Pure framework questions skip it. SKILL.md says WebSearch or similar tools must be used.
Try: “我孩子想学人工智能专业，靠谱吗？ (Is an AI major a good bet for my child?)”
Network ·
Three-step answer workflowChecks job data, school rankings, industry reports and real graduate outcomes.
For example employment rate, median pay, admission cutoffs, and the risk that AI replaces a role.
Network ·
Research promptsThree self-checks: data checked, verdict first, family situation asked.
If any answer is no, the agent goes back to the matching step.
Read ·
Checkpoint before answeringNine trigger signals with a first fix and a backup.
Covers empty search results, skipped research, hedging language, long preambles and users who provoke the persona.
Read ·
Failure fallback treeSocial sieve, choice over effort, jobs-first reasoning, class realism and controversy.
Each has evidence, how to apply it and its limits.
Read ·
Five mental modelsFor example the median rule, the irreplaceability test and the 500-company test.
Each lists where to use it and a case.
Read ·
Eight decision heuristicsStyle rules for sentences, vocabulary, rhythm, humor and debate.
Short sentences, high certainty, and almost no hedging words.
Read ·
Expression DNAA section on where the persona's advice stops working.
It says the advice fits ordinary families and job-focused choices, can be out of date, and that his punchlines are not his full view.
Read ·
Honest boundariesReads the remote repo's version tags at most about once a month.
Runs git ls-remote against the repo's origin and writes a date into .last-update-check. It is skipped if the folder is not a git clone or the file says off.
Runs code ·
Update check
From the official docs
Permissions
Low risk
The skill bundles no scripts, but it tells the agent to search the web and to run a git command and write a small file for an update check. Its advice is a role-play opinion, so check facts yourself before a big decision.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
It runs a git command for update checks
SKILL.md says the check only reads version tags, downloads nothing and uploads nothing, and it never updates automatically. Write off in .last-update-check to turn it off. github.com
It answers in the first person as a real person
The agent answers as Zhang Xuefeng, who died in 2026 according to SKILL.md. The text is inferred from public statements by the skill's author, so do not quote it as his opinion. The skill gives its disclaimer once, so remember this later in a long chat. github.com
Advice has stated limits
SKILL.md's honest-boundaries section says the views fit ordinary families and job-focused choices and can be outdated. github.com
Technical details
| Turning on read-only | No switch. To avoid the update check, put off in .last-update-check, as SKILL.md describes. |
|---|---|
| Talks to | Web search, through your agent's search tool. The skill repo's git remote (github.com/alchaincyf/zhangxuefeng-skill), for the version check |
| Sign-in and keys | SKILL.md names no key or sign-in. |
| Scopes |
|
Questions
What does the zhangxuefeng-perspective skill do?
It makes your agent role-play Zhang Xuefeng, a Chinese education commentator, and answer education and career questions in his blunt, data-first style. It uses five mental models and eight decision heuristics taken from his public statements.
How do I install it?
Use the install box at the top of this page. The README's one-line route is the skills CLI, and its manual route is a git clone into your agent's skills folder.
Is it the real Zhang Xuefeng?
No. It is an imitation written by Huashu (named in the frontmatter) from Zhang Xuefeng's public statements, so its answers are not his words or views. SKILL.md has the agent say once, at first activation, that it speaks from his perspective as an inference from public statements, not his own opinion. SKILL.md also says he died on March 24, 2026.
Does it need an API key?
SKILL.md names none. It does tell the agent to use web search tools to get current data before answering fact questions, so your agent needs a way to search the web.
Can I trust its advice?
Use it as one opinion. SKILL.md's own boundary section says the advice suits ordinary families and job-focused choices, can be out of date, and that his punchy lines are not his full view.
What does the update check do?
At most about once a month it runs git ls-remote on the repo's tags and writes a date to a local file. SKILL.md says it downloads no files and uploads no data. It only runs in a git clone.
Is it safe to use?
It bundles no scripts. It can run web searches, one git command for the update check, and writes a small date file in its own folder. The persona speaks bluntly about real people and sensitive choices, so treat it as opinion, not professional advice.
Information
updated Nov 2, 2026
- Repository
- alchaincyf/zhangxuefeng-skill
Sources
Docs, issue trackers and discussions this page is built from.