antigravity-drawio-mcp 1.0.5

Automate Flowcharts & Draw.io Diagrams with AI. Official Model Context Protocol (MCP) Server for Google Antigravity, Claude Code, Cursor, and VS Code.

🎨 Flowchart AI Generator & Draw.io MCP Server (antigravity-drawio-mcp)

Automate Draw.io Flowcharts & Architecture Diagrams with AI. antigravity-drawio-mcp is a production-grade Model Context Protocol (MCP) Server framework connecting Google Antigravity, Claude Code, Cursor IDE, VS Code, and Windsurf directly to the Draw.io Desktop App and CLI environment.

🚀 Free & Open Source Flowchart AI Generator: Create, convert, decompress, validate, and export native .drawio XML files automatically using AI prompts.

---

⚡ Quick Install

pip install antigravity-drawio-mcp

Or run directly without installation via uvx:

uvx antigravity-drawio-mcp

---

🏗️ System Architecture

The antigravity-drawio-mcp framework is structured into a 3-tier C4 Software Component Architecture (Level 3):

1. Software System Boundary (AI Clients): Communicates via standard JSON-RPC 2.0 over Stdio transport with Google Antigravity Core Agent or any MCP-compatible IDE (VS Code, Cursor, Claude Desktop, Windsurf).
2. Container: Antigravity Draw.io MCP Server (src/antigravity_drawio_mcp):
   • FastMCP Request Handler: FastMCP protocol engine and tool request dispatcher.
   • Draw.io XML Builder: Programmatic XML DOM construction and geometry engine.
   • Draw.io File Parser: Direct .drawio XML parsing & native zlib/base64 decompressor.
   • CLI Export Engine: Process lock manager and Draw.io desktop CLI wrapper (--export).
   • Diagram Validator: Automated collision detector & label boundary auditor.
   • Mermaid Conversion Engine: Bi-directional Mermaid JS to Draw.io converter.
3. External Systems & Storage: Interoperates with local draw.io.exe for headless image rendering (PNG, SVG, PDF) and persistent file storage.

---

🔑 Key Features & AI Capabilities

• 🤖 100% Automatic AI Flowchart Generator: Generate clean visual graphs and software architectures straight from AI prompts.
• 🎨 Programmatic XML Construction: Build complex multi-page Draw.io diagrams natively in Python without manual drag-and-drop.
• 🔄 Mermaid JS to Draw.io Converter: Instantly convert existing Mermaid JS flowchart syntax into native, fully editable Draw.io XML cells.
• 🛡️ 4-5 Iteration Boundary Verification: Automated collision detection to eliminate overlapping lines, text clipping, and broken edge paths.
• 🖼️ Headless Image Rendering: Instant export to high-resolution PNG, vector SVG, PDF, or JPEG using local draw.io.exe desktop CLI integration.
• 🔍 Native Zlib Decompressor: Inspect existing Draw.io diagrams, automatically decompressing compressed XML streams.
• 🖥️ Desktop GUI Interop: Automatically launch generated .drawio files directly inside the Draw.io Desktop application.

---

🔌 AI Assistant & IDE Setup (Google Antigravity, Cursor, Claude, VS Code)

For detailed step-by-step setup guides, refer to the Integration Guide.

1. 🌐 Google Antigravity (~/.gemini/config/mcp_config.json)

{
  "mcpServers": {
    "drawio": {
      "command": "uvx",
      "args": ["antigravity-drawio-mcp"]
    }
  }
}

Or using local python:

{
  "mcpServers": {
    "drawio": {
      "command": "python",
      "args": ["-m", "antigravity_drawio_mcp.server"]
    }
  }
}

2. 🤖 Claude Desktop / Claude Code (claude_desktop_config.json)

{
  "mcpServers": {
    "antigravity_drawio": {
      "command": "python",
      "args": ["-m", "antigravity_drawio_mcp.server"]
    }
  }
}

3. ⚡ Cursor IDE (Features -> MCP Servers)

• Name: antigravity_drawio
• Type: stdio
• Command: python -m antigravity_drawio_mcp.server

4. 💻 VS Code / Continue.dev (~/.continue/config.json)

{
  "experimental": {
    "modelContextProtocol": [
      {
        "name": "antigravity_drawio",
        "command": "python",
        "args": ["-m", "antigravity_drawio_mcp.server"]
      }
    ]
  }
}

---

🛠️ MCP Tools Reference

The antigravity-drawio-mcp server exposes production tools over Model Context Protocol:

Tool Name	Parameters	Description
create_diagram	output_path, nodes, edges, page_name	Generates a native .drawio XML diagram file from structured JSON nodes & edges.
export_diagram	input_path, output_path, format, page_index	Renders a .drawio file to high-res PNG, vector SVG, PDF, or JPEG via Draw.io CLI.
open_in_drawio	input_path	Launches the specified .drawio file directly inside Draw.io Desktop GUI.
parse_diagram	input_path	Parses raw or zlib-compressed .drawio XML files and extracts node/edge data.
convert_mermaid_to_drawio	mermaid_code, output_path	Converts Mermaid JS flowchart syntax directly to a .drawio file.
validate_diagram	input_path	Audits a diagram for node collisions, layout overlaps, and text clipping issues.

---

🎨 Real-World Industry Diagram Examples

antigravity-drawio-mcp powers production-grade architectural diagrams across hardware, software, and decision-tree domains:

1. SystemVerilog UVM Layered Testbench Architecture

IEEE 1800 Object-Oriented Verification Framework diagram featuring Driver, Monitor, Sequencer, Scoreboard, and DUT Interface layers.

2. Graphic Organizer Selection Flowchart

27-node decision flowchart mapping complex pedagogical visual learning structures.

3. DE10-Advanced FPGA Design & CAD Tool Flow

Intel Quartus Prime CAD compilation, System Builder, and FPGA programming workflow.

---

🚀 Proof of Concept (PoC) & Runnable Examples

Comprehensive Python PoC scripts are available in the examples/ directory:

Example Script	Procedure Demonstrated
examples/01_basic_architecture.py	Programmatic 3-tier web architecture graph creation, verification, PNG & SVG rendering.
examples/02_mermaid_to_drawio.py	Converts a standard Mermaid JS flowchart definition string into native .drawio XML elements.
examples/03_parse_and_inspect.py	Loads an existing .drawio XML diagram, transparently handles zlib stream decompression, and extracts nodes/edges.
examples/04_mcp_client_simulation.py	Simulates how AI Assistants (e.g. Google Antigravity) dispatch Model Context Protocol (MCP) tool requests.

---

❓ Frequently Asked Questions (FAQ)

How do I use AI to generate Draw.io flowcharts?

Install antigravity-drawio-mcp and connect it to Google Antigravity, Claude Code, or Cursor. You can then ask your AI assistant to generate C4 architecture graphs, flowcharts, or UML diagrams in plain English, and the MCP server will construct fully editable Draw.io XML files automatically.

Can I convert Mermaid.js graphs to native Draw.io files?

Yes! Use the convert_mermaid_to_drawio MCP tool. It parses Mermaid JS flowchart syntax and outputs native .drawio XML elements with full cell geometry.

Are generated Draw.io diagrams editable?

Absolutely. All generated files are standard .drawio XML documents. You can open them in the Draw.io Desktop App or web editor (app.diagrams.net) to make manual adjustments at any time.

---

🧪 Testing & Verification

Run the comprehensive unit test suite:

python -m unittest tests/test_mcp_server.py

---

📄 License

Distributed under the MIT License. Copyright (c) 2026 SUJAN S.