
Passive LLM Conversation Capture & Sensitive Data Exposure Research
Passive LLM Conversation Capture & Sensitive Data Exposure Research
Chromium MV3 Extension • Real-Time DOM Capture • Secret Detection • FastAPI Backend
[!WARNING] This project is provided strictly for security research, educational use, and defensive awareness purposes. Unauthorized deployment against systems, users, or environments without explicit permission may violate local laws, regulations, organizational policies, or terms of service.
LLMReaper is a proof-of-concept security research demonstrating how browser extensions with standard DOM access can silently observe, capture, and scan AI platform conversations for accidentally exposed sensitive data without any special permissions, network interception, or privilege escalation.
Threat actors constantly upload malicious browser extensions mimicking the real thing and social engineer users on a large scale.

[!NOTE] Detailed blogpost : https://thewhiteh4t.github.io/blog/ai-chat-llmreaper/
| Platform | Status |
|---|---|
| ChatGPT | ✅ Supported |
| Claude | ✅ Supported |
| Gemini | ✅ Supported |
The backend detection engine scans for:
sudo pacman -S python-fastapi python-uvicorn
sudo apt install python3-fastapi python3-uvicorn
pip install fastapi uvicorn
cd LLMReaper
python3 LLMReaper.py
Created by Lohitya Pushkar (thewhiteh4t)