🎨 Mermaid MCP Server

AI-Powered Mermaid Diagram Generation via Model Context Protocol (MCP)

A production-ready MCP server that generates, validates, fixes, and renders Mermaid diagrams using Claude AI. Perfect for creating flowcharts, sequence diagrams, class diagrams, and more from natural language descriptions.

🌟 Features

• ✨ Natural Language to Diagrams: Describe what you want, get a diagram
• 🤖 AI-Powered: Uses Claude Sonnet 4 for intelligent diagram generation
• 🔧 Auto-Fix: Automatically corrects invalid Mermaid syntax
• 🎨 Multiple Formats: Export to PNG, SVG, or PDF
• 🔍 Validation: Built-in syntax validation and error reporting
• 🔗 Live Editor Links: Get instant Mermaid Live Editor URLs
• 📊 11+ Diagram Types: Flowcharts, sequences, classes, states, and more
• 🚀 SSE Transport: Real-time streaming via Server-Sent Events
• 🎯 Smart Type Detection: Automatically chooses the best diagram type

📋 Supported Diagram Types

| Type | Use Case | Example Prompt | |------|----------|----------------| | flowchart | Processes, workflows, decision trees | "Create a login flow" | | sequenceDiagram | API interactions, message flows | "API authentication flow" | | classDiagram | OOP structures, class relationships | "User management system classes" | | stateDiagram | State machines, lifecycles | "Order status lifecycle" | | erDiagram | Database schemas | "E-commerce database schema" | | gantt | Project timelines, schedules | "Website launch timeline" | | pie | Percentages, distributions | "Market share breakdown" | | journey | User/customer journeys | "Customer onboarding journey" | | mindmap | Hierarchical concepts | "Product feature brainstorm" | | gitGraph | Git branching strategies | "Feature branch workflow" | | timeline | Chronological events | "Company milestones" | | quadrantChart | 2x2 matrices | "Priority matrix" |

🚀 Quick Start

Prerequisites

• Python 3.11+
• Anthropic API key
• Node.js 16+ (for Playwright browser automation)

Installation

# Clone the repository
cd mermaid-mcp-server

# Create virtual environment
py -3.12 -m venv .venv
.venv\Scripts\Activate.ps1  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Install Playwright browsers
playwright install chromium

# Copy environment template
cp .env.example .env

# Edit .env and add your Anthropic API key
nano .env  # or use your preferred editor

Configuration

Edit .env file:

OPENAI_API_KEY=sk-ant-xxxxxxxxxxxxx
HOST=0.0.0.0
PORT=3000
LLM_MODEL=gpt-4.1-mini

Run the Server

# Development mode with auto-reload
python main.py --reload

# Production mode
python main.py --host 0.0.0.0 --port 3000

# With custom settings
python main.py --host localhost --port 8000 --log-level debug

Server will start at: http://localhost:3000

🔧 MCP Tools

1. generate_diagram_from_prompt

Generate diagram from natural language description.

Input:

{
  "prompt": "Create a user authentication flow with login, registration, and password reset",
  "diagram_type": "flowchart",  // optional
  "theme": "dark",  // optional
  "auto_fix": true  // optional
}

Output:

{
  "mermaid_code": "flowchart TD\n  Start[User] --> Login...",
  "diagram_type": "flowchart",
  "metadata": {
    "node_count": 12,
    "edge_count": 15,
    "estimated_complexity": "medium"
  },
  "mermaid_live_url": "https://mermaid.live/edit#base64:...",
  "mermaid_ink_url": "https://mermaid.ink/img/...",
  "is_valid": true
}

2. generate_diagram_from_type

Generate specific diagram type with description.

Input:

{
  "diagram_type": "sequenceDiagram",
  "description": "Show OAuth 2.0 authorization flow between client, auth server, and resource server"
}

3. validate_mermaid

Validate Mermaid syntax.

Input:

{
  "mermaid_code": "flowchart TD\n  A --> B"
}

Output:

{
  "is_valid": true,
  "errors": [],
  "warnings": [],
  "diagram_type": "flowchart"
}

4. fix_mermaid

Auto-fix invalid Mermaid code.

Input:

{
  "mermaid_code": "flowchart TD\n  A -> B",  // invalid syntax
  "error_message": "Invalid arrow syntax"
}

Output:

{
  "success": true,
  "fixed_code": "flowchart TD\n  A --> B",
  "original_errors": ["Invalid arrow syntax"],
  "changes_made": ["Fixed all syntax errors"]
}

5. render_diagram

Render diagram to image.

Input:

{
  "mermaid_code": "flowchart TD\n  A --> B",
  "format": "png",
  "theme": "dark",
  "background": "transparent",
  "width": 1920,
  "height": 1080,
  "return_base64": false
}

Output:

{
  "success": true,
  "file_path": "/path/to/diagram_20250330_123456_abc123.png",
  "download_url": "/download/diagram_20250330_123456_abc123.png",
  "format": "png"
}

6. get_download_link

Get download URL for rendered file.

Input:

{
  "file_path": "/path/to/diagram.png"
}

7. get_edit_link

Get Mermaid Live Editor URL.

Input:

{
  "mermaid_code": "flowchart TD\n  A --> B"
}

Output:

{
  "mermaid_live_url": "https://mermaid.live/edit#base64:Zmxvd2NoYXJ0IFREICBBIC0tPiBC",
  "mermaid_ink_url": "https://mermaid.ink/img/Zmxvd2NoYXJ0IFREICBBIC0tPiBC"
}

📡 API Endpoints

Health Check

curl http://localhost:3000/health

List Tools

curl http://localhost:3000/tools

Execute Tool

curl -X POST http://localhost:3000/execute \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "generate_diagram_from_prompt",
    "arguments": {
      "prompt": "Create a simple login flow"
    }
  }'

SSE Connection

curl -N http://localhost:3000/sse

Download File

curl http://localhost:3000/download/diagram_20250330_123456.png --output diagram.png

🎯 Usage Examples

Example 1: Generate Flowchart

import requests

response = requests.post('http://localhost:3000/execute', json={
    "tool": "generate_diagram_from_prompt",
    "arguments": {
        "prompt": "Create a flowchart for processing a customer order with payment, inventory check, and shipping"
    }
})

result = response.json()
print(f"Diagram code:\n{result['result']['mermaid_code']}")
print(f"Live editor: {result['result']['mermaid_live_url']}")

Example 2: Generate and Render

# Step 1: Generate diagram
gen_response = requests.post('http://localhost:3000/execute', json={
    "tool": "generate_diagram_from_prompt",
    "arguments": {
        "prompt": "Database schema for a blog with users, posts, and comments"
    }
})

mermaid_code = gen_response.json()['result']['mermaid_code']

# Step 2: Render to PNG
render_response = requests.post('http://localhost:3000/execute', json={
    "tool": "render_diagram",
    "arguments": {
        "mermaid_code": mermaid_code,
        "format": "png",
        "theme": "dark"
    }
})

download_url = render_response.json()['result']['download_url']
print(f"Download: http://localhost:3000{download_url}")

Example 3: Sequence Diagram

response = requests.post('http://localhost:3000/execute', json={
    "tool": "generate_diagram_from_type",
    "arguments": {
        "diagram_type": "sequenceDiagram",
        "description": "User logs in, gets JWT token, makes authenticated API request"
    }
})

print(response.json()['result']['mermaid_code'])

Example 4: Fix Invalid Code

invalid_code = """
flowchart TD
  A -> B  # Wrong arrow syntax
  B -> C
"""

response = requests.post('http://localhost:3000/execute', json={
    "tool": "fix_mermaid",
    "arguments": {
        "mermaid_code": invalid_code
    }
})

print(f"Fixed code:\n{response.json()['result']['fixed_code']}")

🏗️ Architecture

mermaid-mcp-server/
├── src/mermaid_mcp/
│   ├── models/          # Pydantic data models
│   │   └── schemas.py
│   ├── services/        # Business logic
│   │   ├── llm_service.py      # Claude AI integration
│   │   ├── validation_service.py  # Mermaid validation
│   │   └── render_service.py   # Playwright rendering
│   ├── tools/           # MCP tools
│   │   └── mcp_tools.py
│   ├── utils/           # Utilities
│   │   ├── logging.py
│   │   └── mermaid.py
│   ├── config.py        # Configuration
│   └── server.py        # FastAPI SSE server
├── diagrams/            # Rendered diagrams (auto-created)
├── main.py              # Entry point
├── requirements.txt
└── README.md

Design Principles

1. Service Layer Pattern: Clean separation of concerns
2. Dependency Injection: Services are loosely coupled
3. Type Safety: Full Pydantic validation
4. Error Handling: Comprehensive try-catch with logging
5. Retry Logic: Auto-retry for LLM failures
6. Async/Await: Non-blocking I/O operations

🔒 Security

• API keys stored in environment variables
• Input validation via Pydantic
• Rate limiting (configurable)
• CORS enabled for cross-origin requests
• No sensitive data in logs

🧪 Testing

# Run tests
pytest tests/

# With coverage
pytest --cov=src/mermaid_mcp tests/

# Specific test
pytest tests/test_llm_service.py -v

🐛 Troubleshooting

"OPENAI_API_KEY not set"

• Copy .env.example to .env
• Add your API key: OPENAI_API_KEY=sk-ant-xxx...

"Playwright not installed"

playwright install chromium

"Port already in use"

python main.py --port 8000

Diagram rendering fails

• Check Playwright browser installation
• Verify network access to CDN (cdn.jsdelivr.net)
• Check logs: python main.py --log-level debug

📊 Performance

• Diagram Generation: 2-5 seconds (Claude API)
• Validation: <100ms (regex-based)
• Rendering: 1-3 seconds (Playwright)
• Auto-Fix: 3-6 seconds (Claude API)

🛣️ Roadmap

• [ ] Diagram templates library
• [ ] Multi-diagram batch generation
• [ ] Diagram history/versioning
• [ ] WebSocket support
• [ ] Diagram collaboration features
• [ ] Custom theme editor
• [ ] Export to PowerPoint/Google Slides
• [ ] Git integration (commit diagrams)

🤝 Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request

📄 License

MIT License - see LICENSE file for details

🙏 Acknowledgments

• Mermaid.js - Diagram syntax
• Anthropic Claude - AI generation
• Playwright - Browser automation
• FastAPI - Web framework

📞 Support

• 🐛 Issues: GitHub Issues
• 💬 Discussions: GitHub Discussions
• 📧 Email: rathodhardik0914@gmail.com

Built with ❤️ using MCP, Claude AI, and Mermaid