Mermaid MCP

What Narasimhaponnada/mermaid-mcp MCP server does

Narasimhaponnada/mermaid-mcp MCP server connects AI assistants to Mermaid diagram generation. An assistant can accept a plain-language request such as a microservices architecture, CI/CD workflow, or OAuth authentication sequence and produce a corresponding diagram through MCP.

The project covers more than 22 diagram types. These include flowcharts and block diagrams, sequence diagrams, user journeys, timelines, class diagrams, ER diagrams, C4 diagrams, architecture diagrams, state diagrams, mindmaps, Git flow diagrams, and several data-visualization formats. It also includes more than 50 templates for architecture patterns, workflows, and data models.

How it works

The server runs on Node.js and is implemented in TypeScript. It uses the MCP SDK to expose diagram-generation functionality to compatible clients. Mermaid 10 provides the diagram syntax and rendering layer, while Puppeteer supplies browser-based rendering for consistent output.

Generated output is intended for use in documentation, presentations, browsers, and other technical materials. The repository includes example SVG files covering microservices, cloud infrastructure, CI/CD, data pipelines, and serverless architectures. The project describes its SVG output as XML-validated and suitable for downstream use. Its documented export options include SVG, PNG, and PDF.

Setup and configuration

Narasimhaponnada/mermaid-mcp MCP server requires Node.js 18 or later and npm or yarn. The recommended installation uses the published npm package:

npm install -g @narasimhaponnada/mermaid-mcp-server

After installation, mermaid-mcp --version can be used to check the command. For GitHub Copilot in VS Code, add a server entry whose command is mermaid-mcp. The repository also documents a source-based setup: clone the repository, enter the mermaid-mcp-server directory, install dependencies, and run the build script.

The README identifies GitHub Copilot or Claude Desktop as recommended starting clients. It also documents setup paths for Cursor, direct CLI use, MCP Inspector, and custom applications. No API key, OAuth flow, or external credential is specified in the provided material.

Tools and capabilities

The MCP interface supports natural-language diagram requests from AI assistants and custom MCP applications. Supported diagram families include:

• Process diagrams such as flowcharts and block diagrams
• Interaction diagrams such as sequences, timelines, and user journeys
• Structural diagrams such as classes, entities and relationships, C4, and architecture
• Visualization formats such as pie, XY, Sankey, radar, quadrant, and treemap charts
• Planning and workflow formats such as Gantt, requirements, Kanban, and Git flow

The package is designed for integrations with GitHub Copilot in VS Code, Claude Desktop, Cursor IDE, direct command-line workflows, MCP Inspector, and custom applications. Example prompts in the documentation cover architecture diagrams, GitHub Actions pipelines, and OAuth authentication sequences.

Limitations and notes

The provided material does not list individual MCP tool names or describe a structured input schema, so integrations should be configured through the documented client command rather than relying on undocumented tool identifiers. The server depends on a local Node.js installation and uses browser-based rendering through Puppeteer. Source installation requires a build step, while the global npm installation exposes the mermaid-mcp command directly.

The project is released under the MIT License. The supplied documentation identifies Mermaid, the MCP SDK, and Puppeteer as implementation dependencies or acknowledgments; they are not separate installation targets for this server.