MMCPMarkets
NS

naxiaoduo/scammer.skill

Security testing

This page was translated automatically. Technical details may differ from the original. View the original

81 stars Quality 49 Trend 14MIT
Repository Install

Distill scam playbooks into AI—see which step you’re on. Defense only. Distill scam tricks into AI, see exactly which step you're in—just for precautions

Overview

- What kind of scam is this? - What step are you at now? - What will the scammer say next? - How should you respond (or not) Clone target directory name scammer matches the skill name in SKILL.md; Chinese name, GitHub repository is Scammer.skill. Enter the scam type and sample message as prompted, and Skill will automatically extract the scheme and add it to the local database, making the next detection more accurate. - :scams/example_shazhu/ (pig-butchering scam example), used to compare tactic.md / scripts.md / meta.json formats and directly participate in detection. - :Even if a certain type of scam has not yet been added locally, the signal table in prompts/detection.md can still perform heuristic analysis of common types; To fine-tune a particular type long-term, use /add-scam to create your own entry under ./scams/{slug}/. Samples added by users via /add-scam will expand the local ./scams/, becoming more accurate the more you use them. Running logic: input message → match scam type → locate the current stage → anticipate the next step → provide suggestions - → Automatically extract new tricks → merge into the corresponding scam database without overwriting existing content - → "This is not a scam" / "This is another scam" → Write to the correction layer, effective immediately - → Automatically archive with each update, supports rollback - This tool is for **fraud only, does not provide any scam operation guidance (see docs/COMPLIANCE.md) - Detection results are for reference only; if you have suffered losses, please call immediately (National Anti-Fraud Hotline) - Raw material quality determines recognition accuracy: complete chat records > single message screenshots - this project maintains independently; Usage is based on this README, SKILL.md, and INSTALL.md

Overview (machine translated)

Liar.skill

First Version (v1.0)

"You might not answer the call from the National Anti-Fraud Center, but you can equip this skill first. "

**Distill scam tricks into AI Skills and let it see through every step for you. **

Received a suspicious message? Stick in and tell you in 3 seconds:

  • What kind of scam is this?
  • What step are you at now?
  • What will the scammer say next?
  • How should you reply (or not reply)

Install · Use · Sample Effects · Built-in Scam Library · English

Example of the effect

Scenario One: Impersonating Public Security, Procuratorate, and Court

Scenario 2: Pig Slaughtering Scam

Scenario 3: Online Order Brushing

Installation

Claude Code

Clone the target directory name scammer matches the skill name in SKILL.md; Chinese name 骗子.skill, GitHub repository is Scammer.skill.

OpenClaw

For more detailed explanations, see INSTALL.md.

Dependency (Optional)

Usage

Quick Detection (Most Commonly Used)

In Claude Code:

Or directly:

Add new scam samples

Enter the scam type and sample message as prompted, and Skill will automatically extract the scheme and add it to the local database, making the next detection more accurate.

Administrative Commands

Command Instructions
/detect-scam Detecting a suspicious message
/list-scams List the local scam database
/add-scam Add new scam samples
/scam-stats View scam database statistics
/scam-rollback {slug} {version} Roll back version of a certain scam database
/scam-versions {slug} List the historical archives of a certain slug
/delete-scam {slug} Delete the local scam directory (user confirmation required; irreversible)

Built-in scam database

  • Complete example attached to the warehouse: scams/example_shazhu/ (pig butchering tray example), used to compare tactic.md / scripts.md / meta.json formats, and can directly participate in inspection.
  • General Checking: Even if a certain type of scam has not been added locally, the signal table in prompts/detection.md can still perform heuristic analysis of common types; For long-term fine-tuning of a particular type, please use /add-scam to create custom entries under ./scams/{slug}/.

The table below shows Quick Reference for Common Types (does not mean all warehouses have the corresponding slugs; please refer to the scams/ directory):

Common Types Typical Signal Keywords
Impersonating public security, prosecution, and court Police/prosecutor/secure account/involved
Pig-butchering scheme High income from outside the city / Investment platform / Emotional warm-up

This page was translated automatically. Technical details may differ from the original.

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add naxiaoduo/scammer.skill
New to MCP? Start with the mcpskill.org guide →