ai_bot_detection_poc

psiu/ai_bot_detection_poc

3.2

If you are the rightful owner of ai_bot_detection_poc and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

This document provides a structured summary of a Model Context Protocol (MCP) server designed for detecting bot attacks on social media platforms.

Tools
3
Resources
0
Prompts
0

Social Media Fraud Detection Agent (PoC)

Overview

This is an advanced AI-Powered Forensics Dashboard designed to detect and analyze social media bot networks. It combines a FastAPI/SQLite backend with a React Frontend and a Gemini 2.5 Flash Agent to provide autonomous investigation capabilities.

Key Features

🕵️ Autonmous Agent

  • Self-Correcting Investigator: Can investigate vague queries like "What happened during the spike?" by autonomously checking video stats first.
  • Tools: get_video_stats (Peak Detection), run_read_only_sql (Generic Queries), fetch_suspicious_users.

📊 Forensics Dashboard (React)

  • User Risk Explorer: Sortable list of flagged users with "Alert Reasons" (e.g., Spike Participation).
  • Interactive Charts: Click on any data point in the activity graph to drill down into that specific hour's traffic.
  • Rich User Profiles: Modals displaying simulated Bio, Location, Avatar, and a Risk Narrative explaining why the user was flagged.

🛡️ Detection Logic

  • Fresh Bots: Accounts < 48h old with high velocity.
  • Sleeper Bots: Old accounts dormant for > 90 days that suddenly activate during an attack.
  • Spike Analysis: Statistical anomaly detection on hourly traffic.

Tech Stack

  • Frontend: React (Vite), Chart.js, Lucide Icons.
  • Backend: Python FastAPI, SQLite.
  • AI: Google Gemini 2.5 Flash (via google-genai SDK).

Setup & Run

  1. Environment: Create a .env file:

    GOOGLE_API_KEY=your_gemini_api_key

  2. Install Dependencies:

    # Backend
pip install fastapi uvicorn google-genai python-dotenv asyncio

# Frontend (in /web_app)
cd web_app
npm install

  3. Generate Data:

    python data_gen.py

  4. Launch System:

    • Backend: python web_server.py (Runs on port 8000)
    • Frontend: npm run dev (Runs on port 5173)

  5. Access: Open http://localhost:5173 in your browser.

Project Structure

  • web_server.py: Main FastAPI application and Agent definition.
  • data_gen.py: Generates the mock database with organic vs. bot traffic.
  • web_app/: Source code for the React dashboard.