black-hak/whatsapp-ai-moderator — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a WhatsApp group moderation bot that keeps community chats free from vendors and spam. It should connect through WhatsApp Web with a dedicated number, scan a QR code on first launch, and then watch multiple groups at the same time.

The bot needs to detect commercial spam in text and images. It should catch things like remittance ads, money transfer offers, product sales, flight or tour promos, business links, price lists, flyers, and catalogs. Use AI so it can understand the difference between someone asking where to buy something and someone advertising a business.

When it finds a violation, it should wait a realistic amount of time, delete the bad message, warn the person, and remove repeat offenders depending on the mode. Add strict, moderate, and observer modes. Admins should be able to use commands like ping, help, ban, whitelist, reset warnings, add blacklist words, and change mode. Admins should also be able to DM the bot for private sanction history.

Use Node.js, keep the setup simple with environment variables for the AI keys, and include clear run instructions.

Want more depth? Deep Reverse