mcpField
D
drawio-mcp

drawio-mcp

Local
github.com/yohasacuraMedia & Design5Updated 37w ago

About drawio-mcp

drawio-mcp is an MCP server for generating and editing draw.io / diagrams.net XML diagrams. It serves AI clients that need to create or revise diagrams, with tools for varied diagram types, styling, layout, and routing.

Key features

  1. 01Generate varied diagram types
  2. 02Apply shape, edge, and theme styles
  3. 03Auto-layout and route edges
  4. 04Edit and inspect diagram elements
  5. 05Manage pages and import or export files

Use cases

  1. 01Create a login flowchart from a description
  2. 02Draw UML, ER, or C4 diagrams
  3. 03Revise an existing .drawio file
  4. 04Polish layouts and resolve overlaps or labels

MCP support

The MCP server exposes tools for diagram lifecycle management, drawing, styling, layout, and inspection. MCP clients can create, edit, import, and save diagrams.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"yohasacura-drawio": {
"command": "uvx",
"args": [
"drawio-mcp"
]
}
}
}
Run commanduvx drawio-mcp

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
pypidrawio-mcp1.0.2

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

🎨 Draw.io MCP Server

CI PyPI Python License: MIT

Let your AI assistant create beautiful draw.io diagrams — just describe what you want.

An MCP server that generates draw.io / diagrams.net XML files programmatically. Works with GitHub Copilot (agent mode), Claude Desktop, Cursor, and any MCP-compatible client.


✨ What Can It Do?

CapabilityDetails
🏗️ Diagram typesFlowcharts, UML, ER, C4, BPMN, mindmaps, architecture, network, Kubernetes, and more
🎨 Styling310+ shape presets, 44 edge styles, 21 color themes
📐 Auto-layoutSugiyama DAG, tree, horizontal, vertical, grid, flowchart engines
🔀 Smart routingObstacle-aware orthogonal edge routing with automatic port distribution
✏️ Full editingAdd, move, resize, delete, restyle any element after creation
📄 Multi-pageMultiple pages and layers for complex diagrams
📥 Import/ExportLoad existing .drawio files, modify, and save back
🧹 One-click cleanupPolish command that auto-fixes layout, overlaps, edges, and labels

🚀 Quick Start

Prerequisites
  • Python 3.11+
  • uv (recommended) or pip
Install from PyPI
pip install drawio-mcp

Or with uv:

uvx drawio-mcp
Install from source
git clone https://github.com/yohasacura/drawio-mcp.git
cd drawio-mcp
uv sync
uv run drawio-mcp

🔌 Setup with Your AI Client

VS Code (GitHub Copilot)

Add to your VS Code settings or .vscode/mcp.json:

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

Then use Copilot in Agent mode and ask it to create diagrams.

Claude Desktop

Add to your claude_desktop_config.json:

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

Add to your Cursor MCP settings:

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

💬 Example Prompts

Once connected, try these in your AI chat:

PromptWhat you get
"Create a flowchart for a user login process"Login flow with decision nodes and error paths
"Create a C4 context diagram for an e-commerce system"System context with actors and external systems
"Make a UML class diagram with User, Order, and Product"Class diagram with relationships
"Create an ER diagram for a blog database"Entity-relationship diagram with tables
"Build an org chart for a startup"Tree layout with roles and hierarchy
"Draw an AWS architecture with API Gateway, Lambda, DynamoDB"Cloud architecture diagram
"Create a Kubernetes deployment diagram"K8s diagram with pods, services, ingress
"Make a network topology with servers and firewalls"Network diagram with device icons

💡 Tip: Always end with "...and save it as filename.drawio" to get a file you can open in draw.io.


🛠️ Tool Reference

The server exposes 5 tools, each with an action parameter:

diagram — Lifecycle

create · save · load · import_xml · list · get_xml · add_page · add_layer

draw — Content

add_vertices · add_edges · add_group · update_cells · delete_cells · add_title · add_legend · build_dag · build_full

style — Appearance

build · apply_theme · list_vertex_presets · list_edge_presets · list_themes

layout — Positioning

sugiyama · tree · horizontal · vertical · grid · flowchart · smart_connect · align · distribute · polish · relayout · compact · reroute_edges · resolve_overlaps · fix_labels · optimize_connections · resize_container

inspect — Read-only

cells · overlaps · ports · info


🎨 Style Reference

310+ Vertex Presets (click to expand)

Basic shapes: RECTANGLE ROUNDED_RECTANGLE ELLIPSE CIRCLE DIAMOND TRIANGLE HEXAGON CYLINDER CLOUD PARALLELOGRAM ACTOR PROCESS DOCUMENT DATA_STORE NOTE CARD CALLOUT TEXT CUBE STAR FOLDER and more

Themed boxes: BLUE_BOX GREEN_BOX ORANGE_BOX RED_BOX YELLOW_BOX PURPLE_BOX GRAY_BOX DARK_BOX TEAL_BOX WHITE_BOX

Flowchart: FLOWCHART_PROCESS FLOWCHART_DECISION FLOWCHART_TERMINATOR FLOWCHART_DATA FLOWCHART_DOCUMENT FLOWCHART_DATABASE FLOWCHART_START FLOWCHART_END and 25+ more

UML: UML_CLASS UML_INTERFACE UML_ACTOR UML_COMPONENT UML_PACKAGE UML_NODE UML_LIFELINE UML_FRAME

C4: C4_SYSTEM C4_CONTAINER C4_COMPONENT C4_PERSON C4_EXTERNAL C4_DATABASE C4_WEB_BROWSER

ER: ER_ENTITY ER_WEAK_ENTITY ER_ATTRIBUTE ER_KEY_ATTRIBUTE ER_RELATIONSHIP and more

BPMN: BPMN_TASK BPMN_START_EVENT BPMN_END_EVENT BPMN_GATEWAY BPMN_POOL BPMN_LANE and more

Architecture: ARCH_SERVICE ARCH_DATABASE ARCH_QUEUE ARCH_CLOUD ARCH_API ARCH_GATEWAY ARCH_CONTAINER ARCH_FIREWALL ARCH_SERVER ARCH_LOAD_BALANCER and 20+ more

Network: NETWORK_SERVER NETWORK_ROUTER NETWORK_SWITCH NETWORK_FIREWALL NETWORK_CLOUD NETWORK_STORAGE and 40+ more

Cloud/K8s: AWS_CLOUD AWS_LAMBDA AWS_S3 AWS_EC2 AWS_RDS AWS_SQS K8S_POD K8S_SERVICE K8S_DEPLOYMENT K8S_INGRESS and more

Containers: SWIMLANE GROUP GROUP_TRANSPARENT GROUP_DASHED

Plus: State machine, Sequence diagram, Mindmap, DFD, SysML, ArchiMate, Cisco, Mockup/Wireframe, Infographic, and more

44 Edge Presets (click to expand)

General: DEFAULT ORTHOGONAL STRAIGHT CURVED ENTITY_RELATION DASHED DOTTED BIDIRECTIONAL NO_ARROW

Styled: ROUNDED ROUNDED_DASHED THICK THICK_ROUNDED

Colored: COLORED_BLUE COLORED_GREEN COLORED_RED COLORED_ORANGE COLORED_PURPLE COLORED_YELLOW COLORED_GRAY

UML: UML_ASSOCIATION UML_INHERITANCE UML_IMPLEMENTATION UML_DEPENDENCY UML_AGGREGATION UML_COMPOSITION

ER: ER_ONE_TO_ONE ER_ONE_TO_MANY ER_MANY_TO_MANY ER_ZERO_TO_ONE ER_ZERO_TO_MANY

Flow: DATA_FLOW CONTROL_FLOW ASYNC SYNC BPMN_FLOW BPMN_MESSAGE_FLOW

21 Color Themes (click to expand)

BLUE GREEN YELLOW ORANGE RED PURPLE GRAY PINK TURQUOISE TEAL DARK_BLUE DARK_GREEN DARK_RED DARK_ORANGE DARK_PURPLE DARK WHITE C4_BLUE C4_LIGHT_BLUE C4_SKY C4_GRAY

Connection Ports (click to expand)

TOP BOTTOM LEFT RIGHT TOP_LEFT TOP_RIGHT BOTTOM_LEFT BOTTOM_RIGHT TOP_LEFT_THIRD TOP_RIGHT_THIRD BOTTOM_LEFT_THIRD BOTTOM_RIGHT_THIRD LEFT_TOP_THIRD LEFT_BOTTOM_THIRD RIGHT_TOP_THIRD RIGHT_BOTTOM_THIRD


🧑‍💻 Development

git clone https://github.com/yohasacura/drawio-mcp.git
cd drawio-mcp
uv sync

# Run tests
uv run pytest tests/ -v

# Run the server locally
uv run drawio-mcp
Project Structure
src/drawio_mcp/
├── server.py           # MCP server — 5 tools with 40+ actions
├── models.py           # Core XML model (DrawioFile, Diagram, MxCell)
├── styles.py           # Style builder, presets, and themes
├── layout.py           # Layout helpers (tree, grid, port distribution)
├── layout_engine.py    # Advanced layout (Sugiyama, edge routing, polish)
└── validation.py       # Input validation

📝 Contributing

See CONTRIBUTING.md for guidelines on bug reports, feature requests, and pull requests.

📄 License

MIT © yohasacura

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.