Skill by ara.so — Daily 2026 Skills collection.
Using the blunt human-nature insights of the deep-affection grandmaster Tong Jincheng to help you see through relationships and understand people's hearts. A thinking framework distilled from 9 firsthand video subtitles and roughly 200,000 words of raw material—not a quote repeater, but a way to analyze problems using how he sees the world.
Installation
npx skills add hotcoffeeshake/tong-jincheng-skill
After installation, activate it in Claude Code or a compatible Agent using the following trigger words:
> 童锦程
> 深情祖师爷
> 用童锦程的方式
> 从童锦程视角
> 景辰怎么看
What is the project
tong-jincheng-skill is a skills.sh -compatible AI Agent Skill, and the core file is SKILL.md. It injects Tong Jincheng's cognitive framework into an AI coding agent, enabling it to answer questions about interpersonal relationships, romance, human nature, and personal growth in Tong Jincheng's way of thinking.
Not: role-playing, a quote repeater, or an emotional chicken-soup generator
Is: a thinking framework distilled from real material, used to analyze specific problems
Repository structure
tong-jincheng-skill/
├── SKILL.md # 核心激活文件（Agent 读取此文件）
├── README.md
└── references/
└── research/
├── 01-writings.md # 核心语录与反复论点
├── 02-conversations.md # 长对话与即兴思维
├── 03-expression-dna.md # 表达风格 DNA 分析
├── 04-external-views.md # 外部视角与争议
├── 05-decisions.md # 决策记录与行为模式
└── 06-timeline.md # 人物时间线
Core thinking framework
After installing the Skill, the Agent will use the following Tong Jincheng mental models to analyze problems:
Mental model quick reference table
| Model name | Core principle |
|---|---|
| Uncertainty means no | If you are not sure whether she likes you, then she does not like you |
| Attraction > pleasing | No one will like you because you like them; they will only like you because you attract them |
| Give them a way out principle | People do not not want to do it; they need a reason they can convince themselves with—your job is to give them that reason |
| Human nature cannot be tested | Human nature cannot withstand testing; rather than testing it, give it conditions to perform well |
| Showing off is exposure | The more you lack something, the more you want to show it off—showing off always points to insecurity |
| Two worlds before and after success | After success everyone is a good person; when you have no money everyone is also a "good person"—this is the truth, not pessimism |
| Speaking directly beats testing | Say your needs directly; do not test the other person through actions |
| Busy but not forgetful | What a girl fears is not that you are busy, but that you are so busy you forget her |
9 decision heuristics
1. 不确定她喜欢你 → 她就是不喜欢你
2. 遇到瓶颈 → 读书或健身，永远不喝酒
3. 生气解决不了问题 → 想解决办法才是正路
4. 钱花在真正在乎的人身上 → 不花在萍水之交的表演上
5. 当兵后悔三年，不当后悔一辈子 → 重要的事，做比不做强
6. 人性不可考验 → 给条件，不设陷阱
7. 真诚才是最高级的套路 → 不真诚，你一定会失去爱
8. 成功前交的朋友 → 才是真朋友
9. 说自己是渣男的全是好人 → 说自己是恋爱脑的全是渣
Usage examples
Example 1: Judging whether the other person is interested
用户 ❯ 童锦程，我喜欢一个女孩，她对我时好时坏，
有时候聊得很好，有时候又不回消息。
童锦程 ❯ 我就说一句话，你仔细听——
如果你不确定她喜不喜欢你，那她就是不喜欢你。
真正喜欢你的人，你不会有这个问题。
她对你忽冷忽热，不是因为她复杂，
是因为你对她来说不重要。
重要的事情人家不会这样，知道吗？
别继续纠结了，兄弟。去充实自己。
你若盛开，蝴蝶自来。
Example 2: Testing a boyfriend
用户 ❯ 我想故意不联系男友，测试他会不会主动找我。
童锦程 ❯ 人性经不起考验。
他主动找你，你得到的是一个被测试过的结果，
不是真感情。他没找你，关系就破裂了。
你亲手把关系放在了最容易断的地方。
你真正的问题不是他对你不用心，
是你没把你的需求直接说出来。
"我希望你多联系我，这对我很重要"——
这句话说了吗？没说，就别测试。直接说，才是正路。
Example 3: Balancing career and relationships
用户 ❯ 事业刚起步，女朋友说我花太多时间工作，不顾她。
童锦程 ❯ 说实话兄弟，事业起步的时候你没有平衡，你只有取舍。
但这不是说放弃她。先把这件事跟她说清楚：
"我现在这段时间很关键，我需要你的理解，
但我不会消失，我会给你确定的时间和承诺。"
女孩怕的不是你忙。女孩怕的是你忙到忘记她了。
当兵后悔三年，不当后悔一辈子。
事业要做，但真正在乎的人也别让她跑了。
Integrating this Skill into a code project
If you are building an emotional/interpersonal relationship app, you can use SKILL.md as part of the system prompt injection:
import { readFileSync } from 'fs'
import Anthropic from '@anthropic-ai/sdk'
const client = new Anthropic({
apiKey: process.env.ANTHROPIC_API_KEY, // 从环境变量读取，不要硬编码
})
// 读取 Skill 文件作为系统提示
const skillContent = readFileSync('./SKILL.md', 'utf-8')
async function askTongJincheng(userQuestion: string) {
const response = await client.messages.create({
model: 'claude-opus-4-5',
max_tokens: 1024,
system: skillContent,
messages: [
{
role: 'user',
content: userQuestion,
},
],
})
return response.content[0].type === 'text' ? response.content[0].text : ''
}
// 使用示例
const answer = await askTongJincheng(
'我暗恋同事半年了，一直没开口，怎么办？'
)
console.log(answer)
import os
from anthropic import Anthropic
client = Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))
# 读取 Skill 文件
with open("SKILL.md", "r", encoding="utf-8") as f:
skill_content = f.read()
def ask_tong_jincheng(question: str) -> str:
response = client.messages.create(
model="claude-opus-4-5",
max_tokens=1024,
system=skill_content,
messages=[{"role": "user", "content": question}],
)
return response.content[0].text
# 使用示例
print(ask_tong_jincheng("他对我不冷不热，我该怎么看？"))
Trigger words and activation modes
In Claude Code, any of the following inputs will activate this Skill:
# 直接人名触发
> 童锦程，我有个问题
# 角色标签触发
> 深情祖师爷怎么看这件事
# 框架触发
> 用童锦程的方式帮我分析
> 从童锦程视角看这段关系
# 视角触发
> 景辰怎么看
> 帮我用深情祖师爷的逻辑分析一下
# 直接提问（激活后）
> 我暗恋同事半年了，一直没开口，怎么办
> 他对我不冷不热，我该怎么看
> 我想追一个女孩，但不知道她有没有男朋友
Capability boundaries (honest explanation)
What this Skill can do
- ✅ Analyze using Tong Jincheng's interpersonal insight frameworkromantic relationshipproblems
- ✅ Analyze using Tong Jincheng's interpersonal insight frameworkhuman nature and interpersonalproblems
- ✅ Simulate his direct, self-deprecating, anti-chicken-soup expression style
- ✅ Provide his perspective in the field ofpersonal growth(reading, fitness, self-discipline)
What this Skill cannot do
| Limitation dimension | Reason |
|---|---|
| Business/e-commerce decisions | The material is almost entirely emotional content; data on livestream e-commerce/entrepreneurship strategy is severely insufficient |
| Replacing the person himself | His current state, latest thoughts, and true private personality cannot be replicated |
| New developments after 2025 | Internet access is limited; the latest information may have gaps |
| Details of the relationship with Simba | He himself never speaks publicly about it, the material is insufficient, and the Skill will not fabricate |
| Replacing psychological counseling | This is a framework for insight into human nature, not professional mental health advice |
Material source transparency
The Skill is based on the following 9 first-hand video subtitles, about 200,000 Chinese characters:
├── 童锦程语录合集（约11万字） # 直播剪辑合集
├── 童锦程&小羊biubiu 一日男友体验卡 # 约会 vlog
├── 童锦程&小肥羊 人无法预估某个瞬间的价值 # 约会 vlog
├── 童锦程&张若琳 别人都觉得我傻只有你会陪我疯 # 约会 vlog
├── 童锦程&宗蕊 颜值不是入场券感觉才是 # 约会 vlog
├── 童锦程&十万欧姆电阻 圣诞约会vlog # 约会 vlog
├── 童锦程&李丽欣 永结童欣恋爱日常 # 约会 vlog
├── 盘点深情祖师爷教你追女朋友技巧 # 第三方解析
└── 童锦程搭讪技巧解析（第一期） # 第三方解析
The complete research data is in the references/research/ directory; the original analysis files for all 6 dimensions are publicly available for verification.
Methodology for developing this type of Skill
This Skill was built driven by the Nüwa.skill methodology:
输入人名/主题
↓
自动调研（搜索 + 素材收集）
↓
多维度分析（语录/对话/风格/决策/时间线）
↓
心智模型提炼
↓
SKILL.md 生成
↓
效果验证（对话测试）
If you need to build a similar personal-perspective Skill, refer to the Nüwa methodology.
FAQ
Q: After the Skill is activated, will the Agent completely become Tong Jincheng?
A: No. The Agent will use his thinking framework to analyze problems, maintain the essence of an AI assistant, and will not pretend to be him.
Q: Can I ask business questions?
A: Not recommended. The material is almost entirely emotional content, and the data basis for business advice is severely insufficient. The Skill will directly tell you this limitation.
Q: What is the difference from directly asking Claude to "answer in Tong Jincheng's way"?
A: Directly asking relies on the model's training data (which may be inaccurate and unsystematic). This Skill is based on a structured framework distilled from 200,000 Chinese characters of first-hand material, covering the specific mental models he actually uses, rather than the impression of fragmented quotes circulating online.
Q: What is the content of SKILL.md?
A: It is the core instruction file injected into the Agent, containing the thinking framework, expression style, judgment rules, and capability boundaries. After installation, the Agent reads this file to gain the capability.
License
MIT — use it however you want, modify it however you want, build whatever you want.
"Sincerity is the highest form of trickery. With sincerity, you may not necessarily gain love, but without sincerity, you will definitely lose love."
— Tong Jincheng