docs/PROJECT.md

Mooncrime Project Documentation

Project Overview

Mooncrime is a security research tool designed to analyze screenshots from Shodan, focusing on Industrial Control Systems (ICS) and RTSP camera streams. The tool utilizes the Moondream 3 Vision Language Model (VLM) to provide detailed security analysis of exposed systems.

Goals and Objectives

Primary Goals

1. Security Vulnerability Assessment: Identify and analyze exposed ICS interfaces and camera systems
2. Research Tool Development: Build a reproducible workflow for analyzing internet-connected device screenshots
3. Community Education: Provide a tool for security researchers to understand exposed system risks

Secondary Goals

1. Document common vulnerability patterns in exposed systems
2. Develop best practices for secure system configuration
3. Support academic and professional security research

Technical Architecture

Components

• Moondream 3 Model: Vision Language Model for image analysis
• Daggr Framework: Workflow orchestration and UI generation
• Gradio Interface: User interface for image input and results display
• Hugging Face Spaces: Deployment platform for ZeroGPU inference

Data Flow

1. User provides Shodan screenshot URL or uploads image
2. Image is preprocessed and prepared for analysis
3. Moondream 3 model generates analysis based on prompts
4. Results are formatted and displayed to user

Shodan Query Reference

Industrial Control Systems