claude-mermaid

Claude Mermaid MCP Server

MCP server for rendering Mermaid diagrams in Claude Code with live reload functionality and a built-in skill for expert guidance.

Automatically renders diagrams in your browser with real-time updates as you refine them. Perfect for iterative diagram development and documentation workflows.

!Demo

✨ Features

• 🔄 Live Reload - Diagrams auto-refresh in your browser as you edit
• 🎨 Multiple Save Formats - Export to SVG, PNG, or PDF
• 🌈 Themes - Choose from default, forest, dark, or neutral themes
• 📐 Customizable - Control dimensions, scale, and background colors
• 🪄 Interactive Preview - Pan diagrams by dragging, zoom with browser controls, reset position with one click
• ⬇️ Browser Export - Download diagrams as SVG or PNG directly from the preview
• 🗂️ Multiple Previews - Use preview_id to work on multiple diagrams simultaneously
• 💾 Persistent Working Files - Live previews are stored under ~/.config/claude-mermaid/live
• 🤖 Built-in Skill - Includes a Claude skill with best practices and expert guidance for creating diagrams

Architecture

<details> <summary>Project Architecture Diagram</summary>

!Architecture Diagram

</details>

<details> <summary>User Workflow Diagram</summary>

!User Workflow Diagram

</details>

<details> <summary>Dependencies Diagram</summary>

!Dependencies Diagram

</details>

🚀 Quick Start

1. Install

Plugin Install (Recommended)

In Claude Code, add the marketplace and install the plugin:

/plugin marketplace add veelenga/claude-mermaid
/plugin install claude-mermaid@claude-mermaid

Then restart Claude Code to activate the plugin.

From npm:

npm install -g claude-mermaid

From source:

git clone https://github.com/veelenga/claude-mermaid.git
cd claude-mermaid
npm install && npm run build && npm install -g .

2. Verify Installation

Plugin install: The MCP server is configured automatically. Just verify:

/mcp

You should see mermaid in the MCP server list.

npm install: Configure the MCP server manually:

claude mcp add --scope user mermaid claude-mermaid

Then verify:

claude mcp list

You should see mermaid: claude-mermaid - ✓ Connected

🔌 Other MCP Client Configurations

While this server is optimized for Claude Code, it can work with any MCP-compatible client. Here's how to configure it for other popular tools:

<details> <summary><strong>Codex</strong></summary>

Add to your Codex MCP settings file (~/.codex/mcp_settings.json):

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

Or configure via Codex CLI:

codex mcp add mermaid claude-mermaid

</details>

<details> <summary><strong>Cursor</strong></summary>

Add to your Cursor MCP config file (.cursor/mcp.json or settings):

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

Or use Cursor's settings UI:

1. Open Cursor Settings (Cmd/Ctrl + ,)
2. Navigate to MCP Servers
3. Add a new server with command: claude-mermaid

</details>

<details> <summary><strong>VSCode with Cline Extension</strong></summary>

If using the Cline extension for VSCode:

1. Open VSCode settings (Cmd/Ctrl + ,)
2. Search for "Cline MCP"
3. Add to MCP Settings JSON:

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

</details>

<details> <summary><strong>Windsurf</strong></summary>

Add to Windsurf's MCP configuration file:

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

Configuration location varies by platform:

• macOS: ~/Library/Application Support/Windsurf/mcp.json
• Linux: ~/.config/windsurf/mcp.json
• Windows: %APPDATA%\Windsurf\mcp.json

</details>

<details> <summary><strong>Gemini CLI</strong></summary>

Add to Gemini CLI's MCP configuration file (~/.gemini/mcp.json):

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

Or use the Gemini CLI to configure:

gemini config mcp add mermaid --command claude-mermaid

</details>

<details> <summary><strong>Other MCP Clients</strong></summary>

For any MCP-compatible client, use the standard configuration:

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid"
    }
  }
}

The command claude-mermaid should be available in your PATH after installation.

Note: Some clients may require the full path to the executable:

• Find the path: which claude-mermaid (Unix/macOS) or where claude-mermaid (Windows)
• Use absolute path in config: "command": "/path/to/claude-mermaid"

</details>

💡 Usage

Simply ask Claude Code to create Mermaid diagrams naturally. When installed as a plugin, the built-in mermaid-diagrams skill provides expert guidance, best practices, and automatic workflow management.

Basic Examples

"Create a Mermaid diagram showing the user authentication flow"
"Draw a sequence diagram for the payment process"
"Generate a flowchart for the deployment pipeline"

Advanced Examples

With custom formatting:

"Create a dark theme architecture diagram with transparent background"
"Generate a forest theme flowchart and save to ./docs/flow.svg"

With specific output format:

"Create an ER diagram and save as PDF to ./docs/schema.pdf"
"Save the flowchart as PNG to ./docs/flow.png"

_Note: Browser always shows SVG for live preview, while saving to your chosen format._

Iterative refinement:

"Create a class diagram for the User module"
// Browser opens with live preview
"Add the Address and Order classes with relationships"
// Diagram updates automatically in browser!

Complete Example

"Create a flowchart and save to ./docs/auth-flow.svg:

graph LR
    A[User Login] --> B{Valid Credentials?}
    B -->|Yes| C[Access Granted]
    B -->|No| D[Access Denied]
    C --> E[Dashboard]
    D --> F[Try Again]

    style A fill:#e1f5ff
    style C fill:#d4edda
    style D fill:#f8d7da
"

The diagram will be saved to ./docs/auth-flow.svg and opened in your browser with live reload enabled.

🔧 Tools and Parameters

There are two tools exposed by the MCP server:

1. mermaid_preview — render and open a live preview

• diagram (string, required) — Mermaid diagram code
• preview_id (string, required) — Identifier for this preview session. Use different IDs for multiple concurrent diagrams (e.g., architecture, flow).
• format (string, default svg) — One of svg, png, pdf. Live preview is available only for svg.
• theme (string, default default) — One of default, forest, dark, neutral.
• background (string, default white) — Background color. Examples: transparent, white, #F0F0F0.
• width (number, default 800) — Diagram width in pixels.
• height (number, default 600) — Diagram height in pixels.
• scale (number, default 2) — Scale factor for higher quality output.

1. mermaid_save — save the current live diagram to a path

• save_path (string, required) — Destination path (e.g., ./docs/diagram.svg).
• preview_id (string, required) — Must match the preview_id used in mermaid_preview.
• format (string, default svg) — One of svg, png, pdf. If the live working file for this format doesn’t exist yet, it is rendered on demand before saving.

🎯 How Live Reload Works

1. First render: Opens diagram in browser at http://localhost:3737/{preview_id}
2. Make changes: Edit the diagram through Claude Code
3. Auto-refresh: Browser detects changes via WebSocket and reloads
4. Status indicator: Green dot = connected, Red dot = reconnecting

The live server uses ports 3737-3747 and automatically finds an available port.

Live Preview Controls

• Pan: Click and drag the diagram to move it around
• Zoom: Use browser zoom (Ctrl/Cmd + +/- or pinch-to-zoom on trackpad)
• Reset Position: Click the ⊙ button in the status bar to recenter the diagram
• Export: Click the ⬇ button to download as SVG or PNG

Notes

• Live preview is available for svg format only; PNG/PDF are rendered without live reload.
• For sequence diagrams, Mermaid does not support style directives inside sequenceDiagram.

🖥️ Standalone server

You can start the preview server without an AI agent using the --serve flag:

claude-mermaid --serve

This opens the diagram gallery in your browser with all previously rendered diagrams. Useful for browsing and exporting diagrams outside of a Claude Code session.

🛠️ Development

# Install dependencies
npm install

# Build the project
npm run build

# Run tests
npm test

# Watch mode for development
npm run dev

# Start the MCP server directly
npm start

📝 Troubleshooting

Error: Cannot find package 'puppeteer':

This is a rare environment-specific issue. Try these solutions:

1. Install claude-mermaid globally:

   npm install -g claude-mermaid

1. Reinstall the plugin in Claude Code:

   /plugin uninstall claude-mermaid
   /plugin install claude-mermaid@claude-mermaid

Server not connecting:

# Check if server is installed
claude-mermaid -v

# Reinstall if needed
npm install -g claude-mermaid

# Verify MCP configuration
claude mcp list

Permission denied error:

# Make sure the binary is executable
chmod +x $(which claude-mermaid)

Port already in use:

• The server uses ports 3737-3747
• It will automatically find an available port
• Check if another process is using these ports: lsof -i :3737-3747

Diagrams not rendering or live reload not working:

The server logs to ~/.config/claude-mermaid/logs/:

• mcp.log - Tool requests and diagram rendering
• web.log - HTTP/WebSocket connections and live reload

Enable debug logging in your MCP config:

{
  "mcpServers": {
    "mermaid": {
      "command": "claude-mermaid",
      "env": {
        "CLAUDE_MERMAID_LOG_LEVEL": "DEBUG"
      }
    }
  }
}

Then check the logs:

# View MCP operations
tail -f ~/.config/claude-mermaid/logs/mcp.log

# View WebSocket connections
tail -f ~/.config/claude-mermaid/logs/web.log

Available log levels: DEBUG, INFO (default), WARN, ERROR, OFF

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📄 License

MIT - see LICENSE file for details

🔗 Links

• GitHub Repository
• npm Package
• Claude Code Documentation
• Mermaid Documentation

👀 See Also

If you like this project, you might also be interested in

• preview-skills — preview skills for visualizing files in the browser (markdown, csv, json, mermaid and more)

Hosted deployment

A hosted deployment is available on Fronteir AI.