MCP · Model Context Protocol

What is an MCP server for diagramming?

An MCP server for diagramming is a local bridge that lets AI assistants like Claude and Cursor read, edit, and lay out architecture diagrams programmatically. ArchDraw ships one: instead of describing the diagram you want, the AI manipulates your actual canvas — adding nodes, wiring edges, running layouts, and validating the graph — through a stdio JSON-RPC interface.

The Model Context Protocol (MCP) is an open standard for giving AI models access to tools and data. An MCP server for diagramming exposes diagram operations as tools an AI client can call, so the model can work with a real, structured canvas instead of producing text you have to translate back into a drawing.

From description to real canvas

The AI calls tools to inspect current nodes, add components, and rewire connections — the diagram is edited directly, not recreated.

Runs locally

The server executes on your machine over standard input/output, so canvas data never leaves your computer.

JSON-RPC transport

Requests and responses are structured JSON-RPC messages — fast, debuggable, and standard across MCP clients.

Works with your workflow

Connect from Claude Desktop, Claude Code, or Cursor; the same tools drive generation, layout, validation, and export.

What can the ArchDraw MCP server do?

The server exposes diagram operations as named tools. These cover the full edit loop an AI needs to build and refine an architecture diagram:

generate-diagramCreate a new diagram from a prompt or Mermaid snippet.

update-diagramAdd, remove, or rewire nodes and edges on the active canvas.

validate-diagramCheck the current graph for structural and semantic issues.

fix-layoutRe-run auto-layout to clean up overlapping or cramped nodes.

apply-templateLoad an architecture template onto the canvas.

export-diagramExport the current diagram as JSON, Mermaid, PNG, or SVG.

list-nodesInspect what is currently on the canvas, node by node.

save-checkpoint / load-checkpointSnapshot and restore canvas state during a session.

How to connect it

Add the server to your MCP client configuration (for example ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "archdraw-mcp": {
      "command": "node",
      "args": ["/absolute/path/to/archdraw/mcp-server/dist/index.js"],
      "env": {
        "WORKSPACE_PATH": "/absolute/path/to/diagrams"
      }
    }
  }
}

Full instructions live in the MCP Server Guide, and we wrote a deep dive on the architecture in the engineering blog.

FAQ

Which AI assistants can use the ArchDraw MCP server?

Any MCP-compatible client, including Claude Desktop, Claude Code, Cursor, and other agents that read mcp configuration files.

Is the MCP server secure?

Yes. The server runs locally on your machine and talks to the AI over standard input/output (stdio) using JSON-RPC — no data is sent to a third-party gateway.

Do I need ArchDraw’s cloud service to use the MCP server?

No. The MCP server runs locally and operates on your diagram files directly. It is bundled in the project under the /mcp-server folder and can be run in any Node.js environment.

Why would an AI want to edit a diagram instead of just describing it?

Describing a diagram still leaves you to do the layout. With an MCP server the AI can inspect the actual canvas, place nodes, wire edges, and run layout — so the result is a real, editable diagram you can refine instead of starting from scratch.

Generate a diagram free Read the MCP Server Guide