GitHub - iishyfishyy/mermaid-live-mcp

2 min read Original article ↗

MCP server for generating Mermaid diagrams with live browser preview. Renders diagrams in real-time and supports SVG/PNG export.

npm version

mermaid-live-mcp demo

Install

Claude Desktop

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "mermaid": {
      "command": "npx",
      "args": ["-y", "mermaid-live-mcp"]
    }
  }
}

Cursor

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "mermaid": {
      "command": "npx",
      "args": ["-y", "mermaid-live-mcp"]
    }
  }
}

Claude Code

claude mcp add mermaid -- npx -y mermaid-live-mcp

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "mermaid": {
      "command": "npx",
      "args": ["-y", "mermaid-live-mcp"]
    }
  }
}

Tools

Tool Description
generate_mermaid Generate a diagram from Mermaid syntax and open a live preview in the browser
update_diagram Replace a diagram's Mermaid syntax and re-render the live preview
list_diagrams List all diagrams generated in the current session
export_diagram Write a diagram's SVG to disk

How it works

When you ask your AI assistant to create a diagram, mermaid-live-mcp will:

  1. Parse the Mermaid syntax
  2. Open a browser tab with a live preview
  3. Render the diagram as SVG in real-time via WebSocket
  4. Provide download buttons for SVG and PNG export

Updates to a diagram are pushed instantly to the browser — no page refresh needed.

Supported diagram types

All Mermaid diagram types are supported, including flowcharts, sequence diagrams, class diagrams, state diagrams, ER diagrams, Gantt charts, pie charts, git graphs, mindmaps, timelines, and more.

Development

Project structure

Package Description
packages/mcp-server MCP server (published as mermaid-live-mcp)
packages/preview Live browser preview via WebSocket
packages/core Diagram parsing, layout, and SVG rendering
packages/cli Command-line interface

License

MIT