An AI-powered "Invisible Bodyguard" for Live Streamers. Built for the Ant Media Server AI Hackathon 2026.

Live streamers (coders, gamers, educators) frequently suffer from "accidental doxxing". A single slip-up—flashing an API key, password, or address on screen—can compromise their identity and security in milliseconds.

StreamSentinel is a local AI agent that sits between the streamer and the world. It watches the video feed in real-time using Computer Vision.

1. WATCH: Monitors the RTMP ingest feed frame-by-frame.
2. DETECT: Uses Optical Character Recognition (OCR) to find sensitive text (e.g., "Password", "API Key", "Admin").
3. ACT: Instantly triggers the Ant Media Server REST API to terminate the broadcast before the data propagates.
4. REPORT: Sends a real-time incident report via Discord Webhook to the streamer.

• Streaming Engine: Ant Media Server (Community Edition) running on Docker.
• AI Core: Python 3, OpenCV (Video Processing), EasyOCR (Text Detection).
• Alerting: Discord Webhooks API.
• OS: Arch Linux.

• Docker & Docker Compose
• Python 3.10+
• OBS Studio (for testing)

# We use a custom Docker build for Ant Media Server Community Edition
cd deploy
docker build --network=host -t ant-media-server .
docker run --restart=always -d --network=host --name streamsentinel-core ant-media-server