trap-detector

Trap Detector · Pig-butchering scam detector

Trigger scenarios

• User input contains keywords: friend recommendation, group chat, teacher leading, inside info, must rise, double, surge, guaranteed profit, copy trading
• Explicit request: check it, is it a pig-butchering scam, is it safe, was I scammed
• /scan-trap command

8-signal scan checklist

1. A large number of low-quality accounts recommending it at the same time — web search the stock name + "recommendation" to see whether there are homogenized low-engagement accounts
2. Templated recommendation scripts — search keywords such as "{name} about to explode", "main funds finished accumulating", "target doubles"
3. Paid community / VIP livestream funneling — search "{name} WeChat group", "{name} livestream", "{name} VIP"
4. Fundamentals disconnected from hype — call fetch_financials + fetch_sentiment to compare
5. Abnormal K-line coordination — call fetch_kline to see whether it had already risen sharply before the dense recommendation period
6. Teacher / stock-guru persona promotion — search "{name} teacher", "{name} stock guru"
7. Cross-platform coordinated promotion — search multiple platforms: Xiaohongshu / Douyin / Bilibili / Zhihu
8. Fake research reports / forged news — search "{name} rumor", "{name} debunk", "{name} fake"

Risk rating

Number of signals hit	Rating	Recommendation
0-1	🟢 Safe	Data is normal, no signs of abnormal promotion found
2-3	🟡 Caution	A small amount of promotion; verify the information source
4-5	🟠 Warning	Multiple promotion signals,strongly recommend caution
6+	🔴 Highly suspicious	strongly recommend avoiding, suspected pig-butchering scam characteristics

User keyword weighting

If the user mentions the following words,the signal severity automatically increases by 1 level：

• "a friend recommended it to me" / "someone in the group said" / "a teacher is leading me" → +1
• "inside information" / "guaranteed profit with no loss" → +2
• "must rise" / "double" / "surge" → +1

Output format

{
  "ticker": "...",
  "trap_score": 1-10,           // 反向，越高越安全
  "trap_level": "🟢 安全",
  "signals_hit": [
    {
      "id": 3,
      "name": "付费社群引流",
      "evidence": "搜索发现 5 个公众号同期推送 VIP 群入口",
      "severity": "high",
      "sources": ["url1", "url2"]
    }
  ],
  "user_keyword_boost": 1,
  "recommendation": "🟢 未发现明显推广痕迹，可正常分析。但任何投资都需自己判断。",
  "warning_phrases": ["..."]    // 给用户的明确警告（如有）
}

Check complete

• For each of the 8 signals, give "hit / not hit / insufficient data"
• At least 1 specific evidence URL (unless all are safe)
• recommendation is required and cannot be empty
• If ≥ 4 signals, it must start with "strongly recommend caution" or "strongly recommend avoiding"