Colleague, Together We Are Tokens, Apart We Are Skills: When Companies Start "Distilling" Workers
> "Turn the coldness of parting into a warm Skill. Welcome to cyber-immortality." > — colleague-skill GitHub README
That slogan is dark humor at its finest. A Chinese engineer uploaded an open-source project to GitHub: feed it a coworker's chat logs, work documents, and decision-making habits, and an AI generates a "digital colleague" — one that speaks in their tone, writes code to their standards, and has even learned their blame-shifting style. Within five days of launch, the project earned 6,600+ stars.
But the joke quickly soured. More and more companies began requiring employees to hand over their Skills — distilling years of accumulated tacit knowledge, intuitive judgment, and communication rhythms into replicable, standardized scripts. Workers realized they were personally writing the manual for replacing themselves.
And then, the resistance began.
---
1. From "colleague-skill" to "employee distillation": a workplace horror story
colleague-skill is conceptually simple: use a large language model to extract a person's work style, technical conventions, and decision-making patterns, then package them as a reusable AI Skill. Developer @titanwings wrote a candid README: provide Feishu messages, DingTalk docs, and work emails, and the AI generates your "cyber double."
It was originally just an in-joke in tech circles — until employers discovered its "business value."
In early 2026, the open-source AI framework OpenClaw (nicknamed "lobster") grew explosively. Unlike traditional chatbots, it can directly take over a computer — reading files, operating browsers, running scripts, and decomposing complex tasks. From March onward, big tech moved fast:
- Tencent launched the AI Skills community SkillHub, aggregating 13,000+ Skills
- Alibaba released an enterprise AI work platform, "Wukong," aiming to build the world's largest B2B Skill marketplace
- ByteDance Volcano Engine launched the Find Skill marketplace, packaging industry know-how into freely combinable skill packs
- Mouse trajectories and click positions
- Keystroke content (later changed to "activity summaries")
- Periodic screen captures
- Coverage of 200+ apps and websites: Gmail, GChat, VSCode, and the internal tool Metamate
- Top internal comment: "This makes me deeply uncomfortable. How do we opt out?"
- Bosworth's reply: "On company-issued laptops, there is no opt-out."
- Followed by a cascade of crying, shocked, and angry emoji.
- Allowing tracking pauses of 30 minutes at a time (auto-resuming afterward)
- Expanding exemptions (employees handling sensitive content or concerned about bandwidth)
- Switching from precise keystrokes to "activity summaries"
- "Redis keys must have TTLs; PRs without them get rejected" → "Cache usage follows team conventions"
- "Don't put HTTP calls inside transactions" → "Transaction boundary design should consider reasonableness"
- "API timeouts must trigger circuit breakers" → "System stability requires holistic consideration"
- Under the Personal Information Protection Law (PIPL), companies training AI on employee data need explicit authorization — not blanket, all-purpose consent
- Tsinghua University professor Chen Tianhao argues that "tacit knowledge" (personally accumulated judgment habits and trade-off criteria) should in principle belong to the worker
- Lawyer Xiao Sa of Dacheng Law Offices: distilling someone's likeness, voice, or work logic requires the individual's written authorization specifying scope, duration, and manner of use
- European privacy advocacy group NOYB notes that MCI, even if only indirectly collecting data from EU employees, may violate GDPR's "purpose limitation" principle
- The Irish Data Protection Commission has been informed
- Italy explicitly bans electronic monitoring for productivity tracking; Germany permits keyloggers only in suspected serious criminal cases
- ex.skill: distills your ex's way of speaking — "a gentle cyber reunion"
- boss.skill: "distill the boss into tokens; keep the liberation of productivity for yourself"
- mentor.skill: simulates a mentor's guidance and PUA style
- nuwa-skill: claims it can distill anyone's thinking — "put Musk, Munger, and Feynman to work for you"
- zhang-xuefeng.skill: a digital double of China's famous grad-school-admissions guru (later accused of infringement)
- Does your core capability depend on specific contexts, interpersonal networks, and in-the-moment judgment?
- Does your value come from creating new things, not reusing old experience?
- Do you possess the lubricant AI can't replicate — winning people over, cross-department coordination, gray-area decision-making?
- colleague-skill GitHub: https://github.com/titanwings/colleague-skill
- anti-distill.skill GitHub: https://github.com/leilei926524-tech/anti-distill
- nuwa-skill GitHub: https://github.com/alchaincyf/nuwa-skill
- Meta MCI internal memo (Caixin, 2026-04-23)
- Meta employee pushback and adjustments (Cailianshe, 2026-06-03)
- Meta 10% layoffs + 7,000 reassigned to AI (Caixin, 2026-05-21)
- PIPL and GDPR compliance analysis (Xiao Sa, Dacheng Law Offices)
- Chen Tianhao (Tsinghua University): tacit knowledge and labor rights
- Ifeoma Ajunwa (Yale University): legal analysis of workplace surveillance
An unspoken rule of "mandatory Skill-writing" spread. One internet-company employee revealed: "Leaders mostly watch two metrics — daily Token consumption and Skill output. Departments have explicit quotas; weekly output is mandatory."
But Meta went even harder.
---
2. Meta's "refining" plan: every keystroke trains the AI
On April 22, 2026, Zuckerberg and CTO Andrew Bosworth sent an internal memo announcing the Model Contribution Initiative (MCI) — tracking software installed on employee computers.
What does it record? Everything.
Meta CTO Andrew Bosworth put it bluntly: "Our vision is that AI does the work, and we direct, review, and help them improve."
Zuckerberg's logic at an all-hands was even more naked: "AI models now learn by watching smart people do things. The average intelligence of this company's employees is significantly higher than the people outsourcing firms would bring in."
Translation: you are not employees — you are training data. And higher-quality data than outsourced labelers.
---
3. The workers strike back: the "anti-distillation" movement
After MCI launched, Meta employees exploded.
Resistance escalated. Employees posted flyers in offices and launched an online petition that gathered 1,500+ signatures. They rebranded Meta as an "employee data extraction factory."
Meta was forced to concede (but only with patches):
The core framework didn't change: MCI keeps running, and employees remain a source of AI training data.
Meanwhile, anti-distill.skill went live on GitHub, earning 2,000+ stars in a week and 6 million+ views across the web. Its developer, @Deng Xiaoxian (an AI product manager with a law school background), built something simple yet savage:
> Replace core knowledge with "correct but useless" filler.
For example:
Three cleaning intensities: light (keeps 80% of the core), medium (60%), heavy (40%). Submit the hollowed-out version to the company; keep the real version for yourself.
Deng Xiaoxian made his stance clear: "This isn't opposing technological progress — it's resisting workers being 'refined.' If technology has the power to alienate us, we also have the right to use technology to protect ourselves."
---
4. Legal boundaries: who owns your work experience?
The core question in this conflict: is the experience and judgment accumulated at work the employee's personal asset, or the company's intellectual property?
Chinese legal perspective:
Meta's GDPR predicament:
Yale professor Ifeoma Ajunwa's observation is sharper still:
> "At the US federal level, there are no restrictions on employee surveillance. Recording employee keystrokes subjects white-collar workers to the kind of real-time monitoring previously faced only by delivery drivers and gig workers."
---
5. The essence of distillation: from "humans doing work" to "humans feeding AI that does the work"
The deeper logic here isn't just a privacy dispute — it's a fundamental restructuring of labor relations.
Traditional skill transmission: master to apprentice; experience flows between people, and workers retain bargaining power.
AI-era "distillation": experience is extracted, packaged, and infinitely copied; workers lose their scarcity. Once your decision logic is digitized, the company no longer needs "you" — just "your Skill file."
Meta has demonstrated the full roadmap:
1. Collect data (MCI tracks employee actions) 2. Train models (AI agents trained on employee data) 3. Layoffs (10% announced in May — 8,000 people — while 7,000 were reassigned to AI training) 4. Replacement (pushing a universal "AI builder" role that dissolves job boundaries)
In late May 2026, Meta CFO Susan Li confirmed a new round of layoffs starting in May. Zuckerberg claimed "no more mass layoffs this year" — but on the condition that AI has already taken over enough of the work.
---
6. Even more absurd spin-offs
After colleague-skill went viral, derivative projects proliferated across GitHub like a virus:
These projects look like jokes, but behind them lies a serious philosophical question: when a person's way of thinking can be fully copied, what remains of "that person"?
---
7. Conclusion: keep something in reserve in an era when everyone gets formatted
Back to the original question: should workers worry about being "distilled"?
The answer: yes — but don't stop at worrying.
Worry, because this is genuinely an asymmetric fight. Companies have compute, data, and legal teams; workers have only their own time and experience. The practical effect of anti-distill.skill is more symbolic than real — companies can simply hire better teams to build "anti-anti-distillation."
But don't stop at worrying, because the way out isn't resisting technology — it's redefining "irreplaceability."
Deng Xiaoxian's insight is precise: "Employees are happy to use AI to be efficient and leave work on time, but not to use it to end their own careers."
The real dividing line isn't "will you be distilled?" but:
Law will lag, technology will evolve, but one basic truth won't change:
> What can be distilled was never "you." What gets distilled is only the residue of your past work. Your future self will always be one step ahead of your past self.
---