l

FMLD Panel

FMLD Panel is a macOS SwiftUI demo application that illustrates the structure of a local fraud‑monitoring dashboard. It combines rule‑based transaction evaluation, local data storage, and optional AI analysis through Ollama to demonstrate how a fraud detection interface could be built.

This project is intended primarily as a developer template and learning reference, not a complete production fraud‑detection system.

---

Overview

FMLD Panel demonstrates how to build a fraud‑analysis interface using a modular Swift architecture. It focuses on:

• transaction monitoring UI
• rule‑based risk evaluation
• configurable detection logic
• local data processing
• optional LLM‑assisted analysis

The project is designed to be easy to explore and extend.

---

Features

Core Functionality

• Transaction Dashboard
  View and manage transactions inside a SwiftUI monitoring panel.

• Rule-Based Risk Engine
  Transactions are evaluated using configurable rules defined in JSON.

• Risk Scoring System
  Basic scoring logic categorizes transactions into low, medium, and high risk.

• Local Data Storage
  Transaction data is stored locally using SQLite.

• BIN Lookup (Basic)
  Simple card issuer lookup functionality.

---

Optional AI Analysis

The project includes optional integration with Ollama for local LLM analysis.

If enabled, the model can provide additional insights about flagged transactions.

Supported models include:

• llama3
• mistral
• any Ollama-compatible local model

AI analysis is optional and not required for running the project.

---

Architecture

The project is structured into modular components.

UI Layer

SwiftUI interface for transaction monitoring and risk visualization.

Data Layer

Local storage and transaction persistence.

Rule Engine

Evaluates transactions against configurable detection rules.

AI Layer (Optional)

Provides LLM-based analysis through Ollama.

---

Project Structure

FMLD Panel
│
├── UI
│   └── SwiftUI monitoring dashboard
│
├── Services
│   ├── RulesEngine
│   ├── LocalMLService
│   ├── OllamaService
│   └── FreeBinDatabase
│
├── Database
│   └── SQLite storage
│
└── Rules
    └── rules_config.json

---

Installation

Requirements

• macOS 14+
• Xcode 15+

Setup

Clone the repository:

git clone https://github.com/0ff-set/FMLD

Open the project in Xcode:

FMLD Panel.xcodeproj

Build and run.

---

Optional: Ollama Setup

Install Ollama:

curl -fsSL https://ollama.ai/install.sh | sh

Download a model:

ollama pull llama3

The application will automatically detect the local Ollama instance.

---

Rule Configuration

Fraud detection rules are stored in:

Rules/rules_config.json

Example rule:

{
  "id": "001",
  "name": "High Amount Detection",
  "category": "amount",
  "priority": 90,
  "isActive": true,
  "conditions": [
    {
      "field": "amount",
      "operator": "greaterThan",
      "value": "10000",
      "dataType": "number"
    }
  ],
  "action": "review"
}

Rules can be modified or extended without changing application code.

---

Risk Levels

Transactions are categorized using a simple scoring system:

Score	Risk Level	Action
0–30	Low	Approve
31–70	Medium	Manual review
71–100	High	Block

---

Intended Use

FMLD Panel is useful for:

• learning SwiftUI architecture
• experimenting with rule engines
• prototyping fraud dashboards
• testing local LLM integrations
• building internal monitoring tools

---

License

Proprietary.

If you reuse parts of this project, attribution is appreciated. """

output_file = "/mnt/data/README.md" pypandoc.convert_text(md_text, "md", format="md", outputfile=output_file, extra_args=['--standalone'])

output_file