Guide · Agent skill
How to use the Trap Detector agent skill (杀猪盘检测器)
Updated · written with AI from the project's own documentation and checked against our catalogue data
Trap Detector (杀猪盘检测器) is an agent skill, a folder with a SKILL.md instruction file that tells an AI agent how to run a specific task. This one checks a stock tip you received from a friend, a group chat, a 'teacher' or social media against 8 warning signs of a pig-butchering style promotion, then gives a traffic-light risk rating. This guide shows you how to install it, how to phrase requests so it triggers, and what its output does and does not mean.
Trap Detector (杀猪盘检测器)wbh604/UZI-Skill · MIT Download the skill Quick start
- 1Get the skill folder. Download the trap-detector ZIP from this page and unzip it so SKILL.md sits in a trap-detector folder.
- 2Place it where your agent looks. Copy the folder to ~/.claude/skills/ for personal use, or to .claude/skills/ in a project, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Confirm the requirements. Make sure your agent supports SKILL.md skills and web search, and that fetch_financials, fetch_sentiment and fetch_kline are available from your setup.
- 4Describe the tip. Write a request that names the stock and where the tip came from, such as a friend, a group chat or a teacher.
- 5Read the rating. Check the signals hit, the traffic-light rating and the evidence URLs in the output.
- 6Verify independently. Open the cited sources and check the company's own information before making any decision.
What the skill makes the agent do
When you mention a stock tip from a friend, a group chat, a 'teacher' or a post you saw on Xiaohongshu or Douyin, the skill tells the agent to screen the stock against 8 signals. These include many similar low-engagement accounts pushing the same stock, templated hype phrases, paid VIP groups or live rooms, and a price run-up before promotion. Several signals rely on web search of Chinese social platforms and news, and some rely on financial data tools.
The result is a rating from 🟢 安全 to 🔴 高度可疑 based on how many signals are hit, with a recommendation and supporting source URLs. The skill is a screening aid. It is not a verdict on whether a stock is a scam, and a 🟢 rating does not mean the stock is safe.
- Scans 8 signals, each marked as hit, not hit, or insufficient data
- Rates risk as 🟢 0-1 hits, 🟡 2-3, 🟠 4-5, 🔴 6+
- Raises severity when you say things like 内幕消息, 稳赚不赔 or 必涨
Install the trap-detector skill
Download the ZIP for skills/trap-detector from the page where you found this guide, then unzip it. The skill folder must contain SKILL.md at its top level, along with the references folder. Put the folder in one of two places: your personal skills directory ~/.claude/skills/, or a project's .claude/skills/ directory if you want it only for that project.
In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: place the trap-detector folder in ~/.claude/skills/
- Project install: place it in .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the skill
~/.claude/skills/trap-detector/SKILL.mdWhat you need before it works
The skill needs an agent that supports SKILL.md skills and web search, because it searches Chinese social platforms and news. It also references three financial data tools: fetch_financials, fetch_sentiment and fetch_kline. These are not included in the provided files, so they must come from the wider repository or your own setup. Without them, the fundamentals and K-line signals may return insufficient data.
The skill is written for A-share stocks and Chinese social media. Its keywords and sources are not tuned for other markets or platforms.
- An agent that supports SKILL.md skills
- Web search capability
- fetch_financials, fetch_sentiment and fetch_kline from the wider repository or your setup
Trigger the skill with natural requests
You do not need a special command. Describe the tip in your own words and include a keyword the skill listens for, or ask directly whether something is a scam. The skill also lists a /scan-trap command as a trigger.
Keywords that trigger it include 朋友推荐, 群里, 老师带, 内幕, 必涨, 翻倍, 暴涨, 稳赚 and 跟单. Direct requests include 检测一下, 是不是杀猪盘, 安不安全 and 被套路了吗.
- 朋友推荐我买这只股票，你看看是不是杀猪盘？
- 群里有人说这只票必涨，检测一下风险
- 老师带我买的这只票安全吗？
- 小红书上看到推荐，内幕消息说要翻倍，帮我看看
/scan-trapRead the output
The skill returns a structured result with the ticker, a trap score from 1 to 10 where a higher number means safer, the trap level, each signal that was hit with its evidence and source URLs, any keyword boost, a recommendation and any warning phrases. If four or more signals are hit, the recommendation must open with 强烈建议谨慎 or 强烈建议回避.
Each signal is judged by the criteria in references/eight-signals.md. For example, signal 2 counts as hit when two or more hype phrases such as 即将爆发, 主力建仓完毕 or 目标翻倍 appear in the results.
- 🟢 安全: 0-1 signals hit
- 🟡 注意: 2-3 signals hit, verify your sources
- 🟠 警惕: 4-5 signals hit, be very cautious
- 🔴 高度可疑: 6 or more signals hit, strongly advised to avoid
Safety notes: scripts and network
The skill runs no scripts. It does use the network, because its checks depend on web search and on financial data tools. Its output is a risk opinion based on what those searches and data return. Search results can miss things, so a clean result is not proof of safety.
The skill does not establish that anyone committed fraud, and it is not a legal finding. If you suspect fraud, treat its output as one input to your own checking.
- No scripts are run
- Network access is used for web search and data tools
- Missed scams are possible, so 🟢 does not mean safe
Tips for better results
Give the stock name or ticker and say where the tip came from, since the source type affects keyword weighting. Ask the agent to show the evidence URLs, then open a few yourself. Check whether the financial data tools are connected, because missing data appears as insufficient data rather than a clean result.
Use the skill as a checklist before acting, not as a final answer. Compare its findings with the company's own filings and with independent news.
- Name the stock and the channel (friend, group, teacher, app)
- Ask for the evidence URLs and check them
- Confirm the data tools are available so signals are not left as insufficient data
What people use it for
- Screening a stock recommended in a WeChat group before buying
- Checking whether a teacher or influencer's pick shows signs of coordinated promotion
- Building a structured pre-trade checklist for 'sure win' tips
- Giving a compliance or analyst team a consistent first-pass signal review of a stock being promoted
Frequently asked questions
Is the trap-detector skill free and what is its licence?
The skill's SKILL.md lists the licence as MIT. The source does not state any price, so do not assume any cost beyond your own agent setup.
What do I need to install trap-detector?
You need an agent that supports SKILL.md skills, web search, and the financial data tools fetch_financials, fetch_sentiment and fetch_kline. The data tools are not included in the provided files and must come from the wider repository or your setup.
Where do I put the trap-detector folder?
Place it in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder. In the Claude apps you can upload it under Settings → Capabilities → Skills.
Does a green 🟢 rating mean a stock is safe?
No. A 🟢 rating means 0-1 signals were hit in the searches and data available. The skill can miss many scams, so it can give false reassurance.
Does this work for US stocks or other markets?
The skill is tuned to A-share stocks and Chinese social media. Its keywords and sources are not designed for non-Chinese markets or platforms.
Does the skill prove that someone committed fraud?
No. It only produces a risk opinion and does not establish that anyone committed fraud in legal terms.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Trap Detector suits Chinese-speaking A-share investors who want a structured checklist before acting on a tip from a group, a friend or a short-video influencer, and analysts doing a first-pass review. It is a poor fit if you want confirmation that a stock is safe, if you trade outside A-shares or Chinese platforms, or if you need a legal finding of fraud. For those cases, use it alongside primary filings and independent research.
Download the skill