A professional, backend-focused project that establishes a secure and extensible foundation for an agentic AI honeypot system.
The system is designed to detect scam intent, support multi-turn agentic conversations, and extract structured scam intelligence using modern LLM tooling.
Online scams such as phishing, impersonation, and payment fraud are increasingly adaptive, often bypassing traditional detection systems.
This project provides a production-ready backend architecture that can be extended with agentic AI capabilities to engage scammers covertly and extract actionable intelligence.
- Provide a secure REST API for message ingestion
- Support agentic, multi-turn conversational workflows
- Enable scalable integration of scam detection and intelligence extraction
- Follow clean, professional software engineering practices
- Python 3.10+
- FastAPI – asynchronous REST API framework
- Pydantic – request and response validation
- LangChain – LLM orchestration
- LangGraph – stateful agentic workflow management
- Large Language Models (LLMs) – configurable (OpenAI / Groq / Gemini)
- Git – version control with conventional commits
- Virtual Environments (
venv) – dependency isolation