FEATURE: Agentic Honey-Pot – Autonomous AI Scam Detection API

Describe the Feature

Propose adding an autonomous AI-powered honeypot API that detects scam messages in real-time, engages scammers using a believable human persona, and extracts actionable intelligence like UPI IDs, phishing links, and bank account details. The API supports multi-turn conversations and returns structured JSON output for evaluation and analytics.

Use Case

This API can be used by developers, cybersecurity researchers, or hackathon participants to simulate realistic scam interactions, test scam detection algorithms, and collect intelligence on phishing attempts. It enables autonomous engagement with malicious actors without exposing real users to risk.

Additional Information

The API is secured using an API key, supports session-based conversations, and can detect scam content using both rule-based checks and AI-based natural language understanding. Responses are realistic and human-like, maintaining safety and avoiding exposure of scam detection logic.

Suggested Tools

Node.js / Next.js API routes – for backend API endpoints

MongoDB / Mongoose – for session and conversation storage

OpenRouter SDK – for AI-powered conversation and scam detection

Regex / AI logic – for extracting UPI IDs, phone numbers, and phishing links

Additional Context or Information

This project was inspired by hackathon challenges like Agentic Honey-Pot – AI Scam Detection, and it’s ready for open-source contribution. It can help the community explore real-world AI security applications while providing a safe platform for automated scam engagement simulations.