drawio
Generate draw.io diagrams programmatically using Python. Creates flowcharts, architecture diagrams, tree structures, network diagrams, and more. Use when the user requests a .drawio file, diagram, flowchart, or visual documentation.
When & Why to Use This Skill
This Claude skill enables the programmatic generation of professional draw.io diagrams using Python. By leveraging the drawpyo library, it automates the creation of complex flowcharts, system architectures, and hierarchical trees, transforming structured data or text descriptions into editable visual documentation for enhanced technical communication and project planning.
Use Cases
- System Architecture Mapping: Automatically generate detailed cloud infrastructure or microservices diagrams from text descriptions or configuration files to visualize system components.
- Business Process Modeling: Convert step-by-step business processes, SOPs, or user journeys into professional flowcharts using integrated Python scripts for better workflow clarity.
- Data Structure Visualization: Transform hierarchical data such as organizational charts, decision trees, or file directory structures into visual tree diagrams with automated layouts.
- Technical Documentation: Streamline the creation of network topologies and UML diagrams directly within the AI interface to produce high-quality assets for READMEs and technical guides.
| name | drawio |
|---|---|
| description | Generate draw.io diagrams programmatically using Python. Creates flowcharts, architecture diagrams, tree structures, network diagrams, and more. Use when the user requests a .drawio file, diagram, flowchart, or visual documentation. |
| license | MIT |
| author | example |
| version | "1.1" |
| dependency-management | uv (PEP 723 inline script metadata) |
Draw.io Diagram Generation
Overview
This skill generates .drawio files using the drawpyo Python library. Draw.io diagrams are XML-based and can be opened in:
- draw.io desktop app
- diagrams.net (web)
- VS Code draw.io extension
Quick Start
All scripts include inline dependency metadata (PEP 723). Use uv run to execute them — dependencies are handled automatically in an isolated environment:
# No installation needed — uv handles dependencies automatically
uv run scripts/create_flowchart.py steps.json /mnt/user-data/outputs/flow.drawio
For custom code, you can also use uv run with inline dependencies:
#!/usr/bin/env python3
# /// script
# requires-python = ">=3.10"
# dependencies = ["drawpyo>=0.2.0"]
# ///
import drawpyo
# Create a file and page
file = drawpyo.File()
file.file_path = "/home/claude"
file.file_name = "diagram.drawio"
page = drawpyo.Page(file=file)
# Add a shape
box = drawpyo.diagram.Object(page=page, value="Hello World")
box.position = (100, 100)
# Save
file.write()
Then run with: uv run my_script.py
Decision Tree
What type of diagram?
├── Tree/Hierarchy (org chart, decision tree, file structure)
│ └── Use TreeDiagram class (auto-layout) — see references/REFERENCE.md
│
├── Flowchart (sequential steps with decisions)
│ └── Use helper script: scripts/create_flowchart.py
│ └── Or write custom code with Object + Edge classes
│
├── Architecture/Network (boxes with connections)
│ └── Write custom code using Object + Edge classes
│ └── See references/REFERENCE.md
│
├── From structured data (CSV, JSON, dict)
│ └── Use helper script: scripts/from_data.py
│
└── Complex/Custom
└── Write custom drawpyo code — see references/REFERENCE.md
Key Concepts
Objects (Shapes)
# Basic rectangle
obj = drawpyo.diagram.Object(page=page, value="Label")
obj.position = (x, y) # Coordinates in pixels
obj.width = 120 # Default: 120
obj.height = 60 # Default: 60
# From draw.io shape library
obj = drawpyo.diagram.object_from_library(
page=page,
library="general", # or "flowchart", "basic", etc.
obj_name="process", # shape name from library
value="Process Step"
)
Edges (Connections)
edge = drawpyo.diagram.Edge(
page=page,
source=obj1,
target=obj2,
label="connects to"
)
Styling
# Apply a style string (same format as draw.io)
obj.apply_style_string(
"rounded=1;whiteSpace=wrap;html=1;"
"fillColor=#dae8fc;strokeColor=#6c8ebf;"
)
Helper Scripts
Run with uv run — dependencies are handled automatically:
| Script | Purpose | Usage |
|---|---|---|
scripts/create_flowchart.py |
Create flowcharts from step definitions | uv run scripts/create_flowchart.py input.json output.drawio |
scripts/create_tree.py |
Create tree diagrams with auto-layout | uv run scripts/create_tree.py input.json output.drawio |
scripts/from_data.py |
Create diagrams from JSON/dict data | uv run scripts/from_data.py input.json output.drawio |
Common Shape Libraries
Use with object_from_library(library=..., obj_name=...):
- general:
rectangle,ellipse,process,diamond,parallelogram,hexagon,triangle,cylinder,cloud,document,note,actor - flowchart:
terminator,process,decision,data,document,predefined_process,stored_data,internal_storage,manual_input,manual_operation - basic:
rectangle,ellipse,rhombus,triangle,pentagon,hexagon,octagon
Output
Always save generated .drawio files to /mnt/user-data/outputs/ and use the present_files tool to share with the user.
Next Steps
- references/REFERENCE.md: Complete API documentation, styling options, all shape libraries
- references/examples.md: Example code for common diagram types
- scripts/: Ready-to-use helper scripts