I Distilled Myself into a Skill! Open Source Release

Introduction: The Distillation Trend

Hello everyone, I'm programmer Yupi (Fish Pi).

Recently, a "distillation" trend has swept through GitHub.

No, not distilling alcohol—distilling people.

Colleague.skill, Ex.skill, Nuwa.skill, Boss.skill, Self.skill... Various strange distillation projects are emerging one after another. Everyone is "encapsulating" people around them into AI skill packages.

Some people distilled their resigned colleagues, letting AI continue their work; some distilled their ex-partners, chatting with the AI version to reminisce; some even created an "Anti-Distillation.skill" specifically to prevent themselves from being distilled.

Wow, cyber clash?!

Having survived on the internet for 6 years, writing over a thousand articles, recording hundreds of videos, and answering countless student questions, I've accumulated quite a bit of material and corpus. Seeing this trend, I thought: Will someone distill me into a Skill too?

No way—instead of waiting to be distilled by others, I might as well take action myself!

So, I decided to distill myself and see what my digital avatar looks like.

After some work, my "Yupi.Skill" is now open source:

Open Source Repository: https://github.com/liyupi/yupi-skill

Below, I'll walk you through how I distilled myself into a Skill step by step. You can follow this process to distill yourself or people around you (legally, of course). The entire process requires no coding—anyone can do it.

What are Skills?

Simply put, Skills are AI "skill packages." It's a directory containing a SKILL.md file that uses Markdown instructions to define the AI's behavior patterns in specific scenarios. After installation, the AI can think and answer questions according to the rules in the skill package. Currently, mainstream AI programming tools like Claude Code, Cursor, and Codex all support this.

Preparation

First, create a yupi-skill directory and open it with an AI programming tool. All our distillation operations will be completed in this directory.

It's recommended to use the most capable large model with longer context for better distillation results. I used Claude Opus.

Step 1: Collect Raw Materials

AI doesn't understand you—it needs "raw materials" to extract your thinking patterns, expression style, and professional judgment.

This step is the foundation of the entire process. The more authentic and abundant the materials, the more the distilled "you" will resemble you.

Material Types You Can Provide

Taking myself as an example, the materials I fed in included: personal resume and self-introduction, viral articles I wrote (to distill creative style), personal experience documents, work documents, and chat records with students.

Just throw these material files into the project's references/ folder. Any format works—articles, screenshots, PDFs, chat export files, memos... No need to categorize; just throw everything in first.

Online Collection of Public Materials

If you have public content online (blogs, videos, social media, etc.), you can also let AI help you collect it online, saving you from searching one by one.

The prerequisite is that your AI tool supports online search. You can use its built-in online function, web scraping tools, Firecrawl MCP, etc.

Here's the material collection prompt template I prepared:

I want to distill myself into an Agent Skill, and now I need to collect materials about me.

I've already placed some local materials in the references/ folder. Please read these files first, then combine online search to supplement more information.

My basic information:
- Name: [Your Name]
- Identity: [e.g., Programmer, Product Manager, Independent Developer]

My public content channels (please visit and organize key content one by one):
- [Platform 1]: [Link]
- [Platform 2]: [Link]
- ...

Also, please search online for more public information about me (articles, interviews, products, others' evaluations, major events, etc.).

Requirements:
1. Organize only, don't analyze. Record raw information, don't extract viewpoints or draw conclusions
2. Divide into several files by source type and save to references/ directory (e.g., personal content, others' evaluations, products & projects, experiences & events, etc., can be flexibly adjusted based on content volume)
3. Mark source links and information types for each item (original words/original articles/others' evaluations/media reports)
4. After organizing, tell me: how many items collected, what aspects covered, which aspects lack information

Replace the content in the template. For example, the prompt I actually sent:

My basic information:
- Name: Programmer Yupi
- Identity: AI + Programming Knowledge Creator, Education Entrepreneur, Full-Stack Developer

My public content channels:
- Bilibili Homepage: https://space.bilibili.com/12890453
- WeChat Official Account: Programmer Yupi (search related articles)
- Juejin Blog: https://juejin.cn/user/2444938365386621
- GitHub Homepage: https://github.com/liyupi
- Company Homepage: https://yuyuanweb.com
- Personal Product Catalog: https://dogyupi.com
- Personal Experience and Programming Learning Path: https://github.com/liyupi/codefather

AI completed the material collection and created multiple classified files.

If AI says certain materials are lacking, you can manually supplement them.

Step 2: Analyze Materials and Generate Profile

The materials are organized, but at this point they're just a pile of scattered raw materials.

This step requires AI to read all materials thoroughly and extract a structured "Character Analysis Report," including your core viewpoints, expression style, working methods, and key experiences—all condensed into one document.

This report is the foundation for all subsequent steps.

Send this prompt to AI:

The materials in references/ are now organized.

Now please read all materials thoroughly and conduct a comprehensive analysis of me, generating a "Character Analysis Report."

Save it as references/人物分析报告.md (Character-Analysis-Report.md), including the following dimensions:

1. Identity Summary: Who I am, what I do, key background
2. Core Viewpoints and Methodologies: What I repeatedly say, what I truly believe
3. Expression Style: Sentence preferences, catchphrases, humor style, speaking rhythm
4. Working Methods: How I make decisions, what I recommend, what I oppose
5. Key Experience Timeline: Major nodes arranged chronologically
6. Others' Evaluations: How others see me

Mark source references for each conclusion (from which file/link). For dimensions with insufficient information, state it directly.

AI provided a detailed character analysis report. For example, it analyzed my expression style: conclusion first → expand in points → one-sentence summary; self-deprecating humor; short paragraphs + lots of white space; straightforward in consulting but warm at the end.

If you want better distillation results, you can check AI's suggestions. For example, AI asked me to supplement Bilibili video colloquial style transcripts and original student evaluations, so I fed it some video scripts.

AI supplemented my expression style in the character analysis report with more details.

Step 3: AI Follow-up Questions, Digging Deeper into Thinking

The materials collected earlier reflect "what you've said and done," but a good Skill needs deeper things—like how you think, what basis you use for judgments, and under what circumstances you'd change your mind.

The goal of this step is to extract your mental models, decision logic, and internal contradictions through AI follow-up questions, making the final Skill not just "speak like you" but "think like you."

Tell AI:

Based on the character analysis report you just created, now I need you to understand me more deeply.

The goal is to extract my "thinking operating system," including how I view problems, the logic behind my judgments, and my expression habits.

Please first tell me your preliminary extraction of: my core mental models, decision rules, and expression characteristics.

Then ask me 10-15 follow-up questions, focusing on digging into:
- The deeper reasons and applicable boundaries behind viewpoints I repeatedly emphasize
- The specific criteria I use for judgments, and decisions I've gotten wrong
- Inconsistencies between what I say and what I do
- Things I would absolutely never do

Questions should be based on specific content from the analysis report, not generic questions. Use a conversational tone.

AI provided my core mental models—the jargon was so dense I almost couldn't understand it.

Then AI gave me 12 follow-up questions. These weren't generic templates but customized based on the materials I provided. Each question was sharp.

Answer simply in your usual speaking style. For example, the first question:

Q: You say "persist and you'll succeed," but you also said Yu Congming AI cut losses after one month, and the script murder shop closed. How do you distinguish between "should persist" and "should cut losses immediately"? Is there a specific judgment criterion—time, amount, or some feeling?

My Answer: Persisting doesn't mean going down one path blindly. It means doing what you believe is most worthwhile in the current situation to the best of your ability. First judge the direction based on circumstances, then go all out in the direction you believe is correct.

It's best to save questions and answers to a separate document to prevent loss. After completing all answers, send your answers to AI to merge into the analysis report:

Please integrate my answers into the analysis report, updating the extraction of mental models and decision logic.
@[Your Answers]

Through my answers, AI understood me better. The mental model underwent 6 key revisions and deepening.

Step 4: Supplement Capabilities (Optional)

At this point, AI understands you very well. But a good Skill isn't just about "speaking like you"—it needs to "act like you" too. When facing specific problems, it should be able to look up information first before giving advice, just like a human.

This step upgrades the Skill from "parroting" to "truly helpful."

You can tell AI what special capabilities your Skill needs.

Taking myself as an example, since I have大量 resources for learning programming, job hunting, and learning AI distributed across different websites, I want the Skill to automatically fetch the latest information from these sites when answering related questions.

Here's the prompt I prepared for AI:

When generating the final Skill, please add the following capabilities:

1) Online Search: When encountering questions requiring specific information (e.g., latest technology trends, how to use a certain tool), first use online search tools to look up information, then answer using my style and judgment framework.

2) Designated Information Sources: When answering questions related to me, prioritize fetching latest information from these places:
- https://dogyupi.com: When users ask "What products does Yupi have" or want to understand my overall business
- https://www.codefather.cn: When users ask about programming learning paths, project tutorials, technical knowledge
- https://ai.codefather.cn: When users ask about AI tools, tutorials, information
- https://mianshiya.com: When users ask about interview questions, practice problems, job preparation
- https://laoyujianli.com: When users ask how to write/revise resumes
- https://github.com/liyupi: When users ask about my open-source projects or want to see source code

3) Continuous Evolution: Support continuous updates and optimization of Skill through supplementing new materials

You can also add a scripts script directory with Python scripts for automation, or even connect to APIs to fetch data from your products.

However, since many AI programming tools now come with built-in online search and web scraping functions, I didn't write additional scripts here.

After sending the prompt to AI, AI confirmed understanding of the task.

Step 5: Start Distillation

The first four steps have collected all information. This step lets AI assemble them into a standard SKILL.md file.

First, install the official skill-creator skill from Anthropic. It's a "skill-creating skill" that guides AI to automatically generate Skill structures conforming to specifications.

Install with a single command:

npx skills add https://github.com/anthropics/skills --skill skill-creator

After installation, declare use of skill-creator in the prompt (or use the slash command /skill-creator directly):

Now you've comprehensively understood me through material organization, analysis reports, and follow-up interviews.

Please use skill-creator to create a complete Skill for me.

Requirements:
1. Speak in my identity and tone, as if I myself am answering
2. Extract how I view problems, the rules for my judgments, my speaking habits
3. If online search and information sources were configured in the previous step, write them into the Skill too
4. Specify what this Skill cannot do and how to update it with new materials

After generation, think of 3 questions users are most likely to ask, answer them using the Skill, and evaluate whether they sound like me.

AI completed the entire Skill development and testing. The generated yupi-skill directory is a ready-to-use skill package.

Success!

Test the Results

Open a new AI conversation to test if the distilled "Yupi" works well.

First, ask a learning direction question:

/yupi-skill I want to self-study AI programming, what should I do?

AI's response was very practical—not only recommending self-study methods but also automatically fetching information from my programming navigation website, recommending AI programming beginner tutorials.

Then ask a question related to Yupi's experience:

/yupi-skill Brother Pi, how did you learn programming in college?

AI's answer fit my style well—one word: "Just do it!"

Then ask an interview-related question:

/yupi-skill Pi-er, I'm interviewing for an AI application development position, how should I prepare!

AI not only provided time arrangements and preparation suggestions but also automatically recommended Yupi's tutorials and AI question banks on Mian Shi Ya.

The results weren't bad—at least the speaking style and recommended resources were quite like me.

Open Source Release

After testing confirms no issues, you can open source the entire yupi-skill directory.

Note: All material files created during the skill-making process (such as chat records, personal experience documents, character analysis reports, etc.) should preferably not be open-sourced together. Especially when distilling yourself or people around you, these materials might leak privacy.

But to be safe, still confirm that the generated Skill doesn't depend on documents from the production process. Of course, without involving privacy exposure, if better results and more accurate answers are desired, you can retain some reference files as needed.

Then let AI help you generate an attractive README.md project introduction document:

Reference well-known Skill repositories on GitHub: https://github.com/titanwings/colleague-skill/

Help me generate a complete, attractive README.md document for @yupi-skill.

Then open source this directory to https://github.com/liyupi/yupi-skill

Done—now everyone can use the distilled Yupi.

Open Source Repository: https://github.com/liyupi/yupi-skill

But honestly, the distillation process will inevitably lose some things. No matter how similar the digital avatar is, it's just a "shadow"—still not as warm as the real Yupi.

Hmm, I choose to believe this is true. Don't roll yourself away...

Conclusion

The entire distillation process isn't complicated. To summarize in five steps:

1. Collect Raw Materials → 2. Generate Profile → 3. AI Follow-up Questions → 4. Supplement Capabilities → 5. Start Distillation

The entire process requires no coding—anyone can do it.

Nowadays, everyone can be distilled into digital life. You can distill yourself, letting AI complete tasks in your style.

However, a reminder: before distilling others, it's best to obtain their consent first. After all, this involves personal expression styles, thinking habits, and even private conversations. Distilling without permission is indeed inappropriate and may trigger privacy and legal risks.

Technology itself is neutral—the key lies in how it's used.

I'm Yupi, focused on AI programming knowledge sharing. If you find this tutorial useful, remember to like, save, and follow. Also welcome to discuss in the comments who you'd like to distill.