RegGuard

AI-Enhanced Financial Marketing Compliance Server using Model Context Protocol (MCP)

RegGuard is a sophisticated MCP (Model Context Protocol) server that provides AI-powered regulatory compliance checking for financial marketing content. It leverages OpenAI's GPT-4o-mini model combined with jurisdiction-specific regulatory knowledge to detect violations across multiple financial markets.

RegGuard acts as an intelligent compliance assistant that:

- 🔍 Analyzes Marketing Content: Uses AI to scan financial marketing materials for regulatory violations
- 🌍 Multi-Jurisdiction Support: Covers Singapore (SG), Hong Kong (HK), UAE (AE), and India (IN) regulations
- ⚡ Real-time Compliance Checking: Integrates seamlessly with AI assistants like Claude and Cursor
- 📋 Automatic Disclaimer Insertion: Intelligently places required regulatory disclaimers
- 📊 Audit Trail Generation: Maintains tamper-proof logs of all compliance checks
- 🎯 Contextual AI Analysis: Goes beyond pattern matching to understand regulatory nuances

🧠 Key Concepts & Educational Resources

Understanding MCP (Model Context Protocol)

- MCP Official Documentation- Learn about the open standard for connecting AI assistants to external systems
- MCP GitHub Repository- Technical specification and examples
- Building MCP Servers - Anthropic Guide- Step-by-step guide to creating MCP servers

- MAS Guidelines on Fair Dealing (Singapore)- Singapore's financial marketing regulations
- SFC Investment Advertisement Guidelines (Hong Kong)- Hong Kong securities marketing rules
- DFSA Conduct Rules (UAE)- Dubai financial services regulations
- SEBI Guidelines (India)- Securities and Exchange Board of India regulations

- OpenAI API Documentation- Understanding GPT-4o-mini integration
- Financial AI Compliance Overview- Industry insights on AI in finance
- RegTech Explained- Introduction to regulatory technology

- Python 3.8+(Download Python)
- OpenAI API Key(Get API Key)
- Git(Install Git)

# Clone the repository git clone https://github.com/your-username/regguard-mcp.git cd regguard-mcp # Create virtual environment (recommended) python -m venv regguard-env source regguard-env/bin/activate # On Windows: regguard-env\Scripts\activate # Install dependencies pip install -r requirements.txt

# Create .env file echo 'OPENAI_API_KEY="your-openai-api-key-here"' > .env

# Test server functionality python test_server.py # Test AI integration python test_ai_client.py # Run example usage python example_usage.py

Install Claude Desktop(Download here)

{ "mcpServers": { "regguard": { "command": "python", "args": ["-m", "src.regguard.server"], "cwd": "/full/path/to/regguard-mcp", "env": { "OPENAI_API_KEY": "your-openai-api-key-here" } } } }

- Start Using:

@regguard Please check this marketing copy for Singapore compliance: "Our investment product guarantees 15% annual returns with zero risk!"

Configure MCP in Workspace: Add to your Cursor workspace settings:

{ "mcp.servers": [ { "name": "regguard", "command": ["python", "-m", "src.regguard.server"], "cwd": "./regguard-mcp" } ] }

- Use in Cursor:

@regguard Analyze this financial ad for Hong Kong compliance violations

from regguard_client import RegGuardClient # Initialize client client = RegGuardClient() client.start_server() # Check compliance result = client.check_compliance( html_content="<p>Guaranteed 20% returns!</p>", jurisdiction="sg" ) print(f"Violations found: {len(result['violations'])}")

1.check_rule_violation- AI Compliance Analysis

Analyzes marketing content for regulatory violations across multiple jurisdictions.

@regguard Check this content for Singapore violations: "Join our exclusive investment club! Guaranteed profits of 25% annually with zero risk to your capital. Limited time offer - only 48 hours remaining!"

- Detailed violation descriptions
- Severity levels (Critical, High, Medium, Low)
- Matched content highlights
- Regulatory references
- Actionable recommendations

2.auto_insert_disclaimer- Smart Disclaimer Placement

Automatically inserts jurisdiction-appropriate disclaimers in optimal locations.

@regguard Add appropriate disclaimers for this Singapore investment ad: <div> <h2>Investment Opportunity</h2> <p>High potential returns available.</p> <button>Invest Now</button> </div>

3.export_audit_trail- Compliance Audit Logs

Generates comprehensive audit reports for compliance teams.

Checks server health and AI capability status.

5.list_supported_markets- Available Jurisdictions

Returns:["sg", "hk", "ae", "in"]

regguard-mcp/ ├── src/regguard/ # Core server code │ ├── server.py # Main MCP server │ ├── rules_engine.py # AI compliance engine │ └── audit_writer.py # Audit trail management ├── rules/ # Jurisdiction-specific rules │ ├── sg.yml # Singapore (MAS) │ ├── hk.yml # Hong Kong (SFC) │ ├── ae.yml # UAE (DFSA) │ └── in.yml # India (SEBI) ├── audits/ # Compliance audit logs ├── tests/ # Test files ├── example_usage.py # Usage examples ├── requirements.txt # Python dependencies └── README.md # This file

# 1. Clone this repository git clone https://github.com/your-username/regguard-mcp.git # 2. Navigate to project directory cd regguard-mcp # 3. Create Python virtual environment python -m venv venv source venv/bin/activate # or venv\Scripts\activate on Windows # 4. Install dependencies pip install -r requirements.txt

# 1. Get OpenAI API key from https://platform.openai.com/api-keys # 2. Create .env file echo 'OPENAI_API_KEY="your-actual-api-key"' > .env # 3. Test configuration python test_server.py

Choose your preferred integration method:

- Configure MCP settings as shown above
- Restart Claude Desktop
- Use@regguardcommands

- Add MCP configuration to workspace
- Restart Cursor
- Use@regguardin your code

- Use the provided example scripts
- Implement JSON-RPC communication
- Follow MCP protocol specification

# 1. Modify jurisdiction rules in rules/ directory # 2. Add custom compliance patterns # 3. Extend supported markets if needed # 4. Customize disclaimer templates

Content: "Guaranteed 25% returns with zero risk!" Expected: Multiple critical violations detected

Content: "Investment involves risk. Past performance is not indicative of future results." Expected: No violations, compliance status PASS

- CRITICAL: Serious regulatory breaches requiring immediate attention
- HIGH: Important compliance issues that should be addressed
- MEDIUM: Moderate concerns that may need review
- LOW: Minor issues or best practice recommendations

- Contextual Understanding: Goes beyond keyword matching
- Regulatory Knowledge: Trained on jurisdiction-specific rules
- Confidence Scoring: AI confidence in each violation detection
- Actionable Recommendations: Specific suggestions for fixes
- Fork the repository
- Create a feature branch (git checkout -b feature/amazing-feature)
- Commit your changes (git commit -m 'Add amazing feature')
- Push to the branch (git push origin feature/amazing-feature)
- Open a Pull Request

This project is licensed under the MIT License - see theLICENSEfile for details.

Global e-invoicing compliance: Factur-X, Peppol UBL, CII & more