README.md

metadata

title: fraudoo
emoji: 🐢
colorFrom: green
colorTo: purple
sdk: gradio
sdk_version: 6.3.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: fraudoo

🛡️ Documentary Fraud Management MCP Service

A robust documentary fraud detection service using LLM Services 3 Flash and Nano Banana Pro, integrated with a Gradio UI and serving as an MCP (Model Context Protocol) server.

✨ Features

• Dual-Model Analysis:
  • LLM Services 3 Flash: Handles complex reasoning, entity extraction (Names, Dates, IDs), and overall fraud scoring.
  • LLM Services Nano Banana Pro: Performs critical visual analysis to detect clones, font mismatches, and retouching.
• Support for Images & PDFs: Automatically converts PDF pages to images for visual manipulation detection.
• Dynamic Context: Prompts include real-time date/time (currently 2026-01-21) for accurate expiration and inconsistency checks.
• Historical Explorer:
  • List View: Browse past analyses with labels, amounts, and fraud scores (X/100).
  • Persistence: Local storage of uploaded files in uploads/ for persistent previews.
  • Deletion: Remove unwanted analysis records and associated raw files.
• Duplicate Detection: Uses SHA-256 for exact matches and ChromaDB for semantic similarity checks.
• Native MCP Server: Exposes analysis tools directly to MCP-compatible clients.

🚀 Getting Started

Prerequisites

• Python 3.10+
• Google API Key (LLM Services)

Installation

1. Clone the repository and navigate to the directory.
2. Create and activate a virtual environment:

   python3 -m venv venv
   source venv/bin/activate
   

3. Install dependencies:

   pip install -r requirements.txt
   

Running the Service

1. Set your Google API Key:

   export GOOGLE_API_KEY="your_api_key_here"
   

2. Enable MCP functionality:

   export GRADIO_MCP_SERVER=True
   

3. Launch the app:

   python app.py
   

4. Open the UI at http://127.0.0.1:7864.

🛠️ Architecture

• app.py: Gradio interface, session management, and persistent file handling.
• fraud_analyzer.py: Core logic for LLM Services integration, PDF conversion, and metadata extraction.
• vector_service.py: Fingerprinting and history storage using ChromaDB.
• uploads/: Persistent storage for analyzed documents.
• chroma_db/: Vector database storage.

🔌 MCP Integration

The Gradio app natively serves as an MCP server. You can access the protocol at: http://127.0.0.1:7864/gradio_api/mcp/