Hello everyone, I'm programmer Yupi.

Recently, GitHub has witnessed a surge of "distillation" projects sweeping through the developer community. And no, this isn't about distilling alcoholic beverages—this is about distilling people into digital personas.

Projects with names like colleague.skill, ex-partner.skill, nuwa.skill (referencing the Chinese goddess who created humanity), boss.skill, and self.skill have emerged one after another. Everyone is busy "encapsulating" people around them into AI skill packages. Some have distilled their departed colleagues, allowing AI to continue doing their work. Others have distilled their ex-partners, chatting with AI versions to reminisce about old times. There's even someone who created an "anti-distillation.skill" specifically designed to prevent themselves from being distilled.

This is genuinely cyber warfare at its finest!

Having survived on the internet for six years, I've written over a thousand articles, recorded hundreds of video episodes, and answered countless questions from students. Along the way, I've accumulated substantial material and language data. Seeing this trend, I started wondering: will someone distill me into a Skill too?

Rather than waiting to be distilled by others, I decided to take matters into my own hands!

So I made the decision to distill myself and see what my digital doppelgänger would look like.

After going through the entire process, my "Yupi.Skill" is now open source and available to everyone.

You can find the open-source project here: https://github.com/liyupi/yupi-skill

Now let me walk you through exactly how I distilled myself into a Skill, step by step. You can follow this same process to distill yourself or people around you (legally and ethically, of course). The entire process requires no coding skills—if you have hands, you can do this.

Understanding Skills

Before diving in, let's clarify what Skills actually are. Simply put, a Skill is an AI "skill package"—a directory containing a SKILL.md file that uses Markdown instructions to define the AI's behavior patterns in specific scenarios. Once installed, the AI thinks and answers questions according to the rules defined in the skill package. Currently, mainstream AI programming tools like Claude Code, Cursor, and Codex all support Skills.

Preparation Work

First, create a new directory called yupi-skill and open it with your AI programming tool. All distillation operations will be completed within this directory.

I recommend using the most capable model with a longer context window for better distillation results. For this project, I used Claude Opus.

Step 1: Collecting Raw Materials

The AI doesn't know you yet. It needs "raw materials" to extract your thinking patterns, expression style, and professional judgment.

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

Here are the types of materials you can provide:

Self-introduction and Resume: Include your name, years of experience, professional background, MBTI personality type, and other identity markers. This helps establish your identity positioning and personality characteristics.

Personal Experiences: Document key stories from your education, career transitions, and major life changes. These reveal your values and growth trajectory.

Chat Records: Export conversations from WeChat, Feishu, DingTalk, or other messaging platforms. These capture your authentic speaking style, catchphrases, and communication patterns.

Work Documents: Include weekly reports, project proposals, code review comments, and professional communications. These demonstrate your professional judgment and working methods.

Creative Content: Gather blog posts, video scripts, social media posts, and朋友圈 (Moments) updates. These showcase your viewpoints and expression style.

Others' Evaluations: Collect feedback from colleagues and friends. These reveal blind spots—characteristics you might not see in yourself.

In my case, I fed in my resume and self-introduction, viral articles I'd written (to distill my creative style), personal experience documents, work documents, and chat records with students.

Simply drop all these material files into the project's references/ folder. Any format works—articles, screenshots, PDFs, exported chat logs, memos. Don't worry about categorization initially; just throw everything in.

Collecting Public Materials via Internet

If you have public content online (blogs, videos, social media), you can let the AI help you collect these materials automatically, saving you from gathering them one by one.

This requires your AI tool to support internet search capabilities. You can use built-in search functions, web scraping tools, Firecrawl MCP, or similar utilities.

Here's the prompt template I prepared for material collection:

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

I've already placed some local materials in the references/ folder. Please read these files first, then supplement with additional information through internet search.

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

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

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

Requirements:
1. Organize only, don't analyze. Record raw information without extracting conclusions
2. Split into several files by source type and save to references/ (e.g., personal content, others' evaluations, products and projects, experiences and events)
3. Mark source links and information types for each item (original quotes/my articles/others' evaluations/media reports)
4. After completion, tell me: how many items collected, what aspects covered, which aspects need more information

After replacing the template content with my actual information, the AI completed the material collection and created multiple categorized files automatically.

If the AI indicates missing materials in certain areas, you can manually supplement them.

Step 2: Analyzing Materials and Generating a Portrait

With materials organized, they're still just a pile of scattered raw ingredients at this point.

This step requires the AI to read through all materials comprehensively and extract a structured "Personal Analysis Report." This document should condense your core viewpoints, expression style, working methods, and key experiences into a single comprehensive document.

This report forms the foundation for all subsequent steps.

Here's the prompt I used:

The materials in references/ are now organized.

Please read through all materials and conduct a comprehensive analysis of me. Generate a "Personal Analysis Report" and save it as references/Personal Analysis Report.md, including the following dimensions:

1. Identity Overview: Who I am, what I do, key background
2. Core Viewpoints and Methodologies: What I repeatedly emphasize, 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 perceive me

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

The AI produced a detailed personal analysis report. For example, it analyzed my expression style as: conclusion first → point-by-point expansion → one-sentence summary; self-deprecating humor; short paragraphs with generous white space; direct but warm conclusions in consultations.

If you want better distillation results, review the AI's suggestions. In my case, the AI recommended supplementing with Bilibili video transcript styles and original student evaluation texts, so I fed it some additional video scripts.

The AI then enriched the personal analysis report with more details about my expression style.

Step 3: AI Follow-up Questions to Excavate Deeper Thinking

The materials collected so far reflect "what you've said and done." However, a good Skill needs something deeper: how you think, what criteria you use for judgment, and under what circumstances you change your mind.

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

Here's the prompt:

Based on the personal analysis report you just created, 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.

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 excavating:
- The deeper reasons and applicable boundaries behind viewpoints I repeatedly emphasize
- My specific criteria for making 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 templates. Use a conversational tone.

The AI provided my core mental models (though the jargon was so dense I barely understood it myself initially).

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

Answer using your normal speaking style. For example, the first question asked:

"You say 'persistence will lead to success,' but you also said Yucongming AI cut losses after one month, and you closed your script murder mystery shop. How do you distinguish between 'should persist' and 'should cut losses immediately'? Is there a specific judgment criterion—time, money, or some feeling?"

My answer: Persistence doesn't mean stubbornly continuing down one path. It means doing your absolute best on what you believe is most worthwhile under current circumstances. First judge the direction based on the situation, then pour your full effort into 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 them to the AI for integration into the analysis report:

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

Based on my answers, the AI gained deeper understanding of me. The mental models section received 6 key corrections and refinements.

Step 4: Supplementing Capabilities (Optional)

At this point, the AI understands you well. However, a good Skill isn't just about "sounding like you"—it needs to "act like you" too. When facing specific problems, it should be able to research information first before giving recommendations, just like a human would.

This step elevates the Skill from "parroting" to "genuinely helpful."

You can tell the AI what special capabilities your Skill needs.

In my case, since I have extensive resources for learning programming, job hunting, and AI distributed across different websites, I wanted the Skill to automatically fetch latest information from these sites when answering related questions.

Here's the prompt I prepared:

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

1) Internet Search: When encountering questions requiring specific information (like latest technology trends, tool usage), first use internet search tools to research, then answer using my style and judgment framework.

2) Designated Information Sources: When answering questions related to me, prioritize fetching latest information from these locations:
- 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, news
- https://mianshiya.com: When users ask about interview questions, practice problems, job preparation
- https://laoyujianli.com: When users ask how to write or revise resumes
- https://github.com/liyupi: When users ask about my open-source projects or want to view source code

3) Continuous Evolution: Support updating and optimizing the Skill through supplementary new materials

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

However, since many AI programming tools now have built-in internet search and web scraping capabilities, I didn't write additional scripts for this project.

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

Step 5: Beginning the Distillation

The first four steps have gathered all necessary information. Now it's time for the AI to assemble everything into a standard SKILL.md file.

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

Installation requires just one command:

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

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

Now you have 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'm answering personally
2. Extract how I view problems, the rules for my judgments, and my speaking habits
3. If internet search and information sources were configured in the previous step, include those in the Skill
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 the responses resemble me.

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

Mission accomplished!

Testing the Results

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

First, I asked a learning direction question:

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

The AI's response was very pragmatic. It not only recommended self-study methods but also automatically fetched information from my programming navigation website, recommending AI programming beginner tutorials.

Next, I asked a question related to Yupi's personal experience:

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

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

Then I asked an interview-related question:

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

The AI not only provided time arrangements and preparation suggestions but also automatically recommended Yupi's tutorials and the AI question bank on Interview Duck.

The results were quite good—at least the speaking style and recommended resources felt very much like me.

Open Source Release

After testing confirmed everything works properly, it was time to open-source the entire yupi-skill directory.

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

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

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

Referencing 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

Now everyone can use the distilled Yupi.

Open-source link: https://github.com/liyupi/yupi-skill

Honestly, the distillation process inevitably loses some things. No matter how similar the digital doppelgänger is, it's still just a "shadow"—it lacks the warmth of the real human Yupi.

Well, I choose to believe that's the case. Otherwise, I might work myself out of existence...

Final Thoughts

The entire distillation process isn't complicated. Summarized in five steps: Collect Materials → Generate Portrait → AI Follow-up Questions → Supplement Capabilities → Begin Distillation.

The entire process requires no coding—if you have hands, you 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 someone without permission is inappropriate and may trigger privacy and legal risks.

Technology itself is neutral. The key lies in how we use it.

I'm Yupi, focused on AI programming knowledge sharing. If you found this tutorial helpful, remember to like, save, and follow. Feel free to share in the comments who you'd like to distill!