Agent skill
draw-io
draw.io diagram creation, editing, and review. Use for .drawio XML editing, PNG conversion, layout adjustment, and AWS icon usage.
Install this agent skill to your Project
npx add-skill https://github.com/davila7/claude-code-templates/tree/main/cli-tool/components/skills/creative-design/draw-io
SKILL.md
draw.io Diagram Skill
1. Basic Rules
- Edit only
.drawiofiles - Do not directly edit
.drawio.pngfiles - Use auto-generated
.drawio.pngby pre-commit hook in slides
2. Font Settings
For diagrams used in Quarto slides,
specify defaultFontFamily in mxGraphModel tag:
<mxGraphModel defaultFontFamily="Noto Sans JP" ...>
Also explicitly specify fontFamily in each text element's style attribute:
style="text;html=1;fontSize=27;fontFamily=Noto Sans JP;"
3. Conversion Commands
See conversion script at scripts/convert-drawio-to-png.sh.
# Convert all .drawio files
mise exec -- pre-commit run --all-files
# Convert specific .drawio file
mise exec -- pre-commit run convert-drawio-to-png --files assets/my-diagram.drawio
# Run script directly (using skill's script)
bash ~/.claude/skills/draw-io/scripts/convert-drawio-to-png.sh assets/diagram1.drawio
Internal command used:
drawio -x -f png -s 2 -t -o output.drawio.png input.drawio
| Option | Description |
|---|---|
-x |
Export mode |
-f png |
PNG format output |
-s 2 |
2x scale (high resolution) |
-t |
Transparent background |
-o |
Output file path |
4. Layout Adjustment
4.1. Coordinate Adjustment Steps
- Open
.drawiofile in text editor (plain XML format) - Find
mxCellfor element to adjust (search byvalueattribute for text) - Adjust coordinates in
mxGeometrytagx: Position from lefty: Position from topwidth: Widthheight: Height
- Run conversion and verify
4.2. Coordinate Calculation
- Element center coordinate =
y + (height / 2) - To align multiple elements, calculate and match center coordinates
5. Design Principles
5.1. Basic Principles
- Clarity: Create simple, visually clean diagrams
- Consistency: Unify colors, fonts, icon sizes, line thickness
- Accuracy: Do not sacrifice accuracy for simplification
5.2. Element Rules
- Label all elements
- Use arrows to indicate direction (prefer 2 unidirectional arrows over bidirectional)
- Use latest official icons
- Add legend to explain custom symbols
5.3. Accessibility
- Ensure sufficient color contrast
- Use patterns in addition to colors
5.4. Progressive Disclosure
Separate complex systems into staged diagrams:
| Diagram Type | Purpose |
|---|---|
| Context Diagram | System overview from external perspective |
| System Diagram | Main components and relationships |
| Component Diagram | Technical details and integration points |
| Deployment Diagram | Infrastructure configuration |
| Data Flow Diagram | Data flow and transformation |
| Sequence Diagram | Time-series interactions |
5.5. Metadata
Include title, description, last updated, author, and version in diagrams.
6. Best Practices
6.1. Background Color
- Remove
background="#ffffff" - Transparent background adapts to various themes
6.2. Font Size
- Use 1.5x standard font size (around 18px) for PDF readability
6.3. Japanese Text Width
- Allow 30-40px per character
- Insufficient width causes unintended line breaks
<!-- For 10-character text, allow 300-400px -->
<mxGeometry x="140" y="60" width="400" height="40" />
6.4. Arrow Placement
- Always place arrows at back (position in XML right after Title)
- Position arrows to avoid overlapping with labels
- Keep arrow start/end at least 20px from label bottom edge
<!-- Title -->
<mxCell id="title" value="..." .../>
<!-- Arrows (back layer) -->
<mxCell id="arrow1" style="edgeStyle=..." .../>
<!-- Other elements (front layer) -->
<mxCell id="box1" .../>
6.5. Arrow Connection to Text Labels
For text elements, exitX/exitY don't work, so use explicit coordinates:
<!-- Good: Explicit coordinates with sourcePoint/targetPoint -->
<mxCell id="arrow" style="..." edge="1" parent="1">
<mxGeometry relative="1" as="geometry">
<mxPoint x="1279" y="500" as="sourcePoint"/>
<mxPoint x="119" y="500" as="targetPoint"/>
<Array as="points">
<mxPoint x="1279" y="560"/>
<mxPoint x="119" y="560"/>
</Array>
</mxGeometry>
</mxCell>
6.6. edgeLabel Offset Adjustment
Adjust offset attribute to distance arrow labels from arrows:
<!-- Place above arrow (negative value to distance) -->
<mxPoint x="0" y="-40" as="offset"/>
<!-- Place below arrow (positive value to distance) -->
<mxPoint x="0" y="40" as="offset"/>
6.7. Remove Unnecessary Elements
- Remove decorative icons irrelevant to context
- Example: If ECR exists, separate Docker icon is unnecessary
6.8. Labels and Headings
- Service name only: 1 line
- Service name + supplementary info: 2 lines with line break
- Redundant notation (e.g., ECR Container Registry): shorten to 1 line
- Use
<br>tag for line breaks
6.9. Background Frame and Internal Element Placement
When placing elements inside background frames (grouping boxes), ensure sufficient margin.
- YOU MUST: Internal elements must have at least 30px margin from frame boundary
- YOU MUST: Account for rounded corners (
rounded=1) and stroke width - YOU MUST: Always visually verify PNG output for overflow
Coordinate calculation verification:
Background frame: y=20, height=400 -> range is y=20-420
Internal element top: frame y + 30 or more (e.g., y=50)
Internal element bottom: frame y + height - 30 or less (e.g., up to y=390)
Bad example (may overflow):
<!-- Background frame -->
<mxCell id="bg" style="rounded=1;strokeWidth=3;...">
<mxGeometry x="500" y="20" width="560" height="400" />
</mxCell>
<!-- Text: y=30 is too close to frame top (y=20) -->
<mxCell id="label" value="Title" style="text;...">
<mxGeometry x="510" y="30" width="540" height="35" />
</mxCell>
Good example (sufficient margin):
<!-- Background frame -->
<mxCell id="bg" style="rounded=1;strokeWidth=3;...">
<mxGeometry x="500" y="20" width="560" height="430" />
</mxCell>
<!-- Text: y=50 is 30px from frame top (y=20) -->
<mxCell id="label" value="Title" style="text;...">
<mxGeometry x="510" y="50" width="540" height="35" />
</mxCell>
7. Reference
- Layout Guidelines
- AWS Icons
- AWS Icon Search Script
AWS icon search examples:
python ~/.claude/skills/draw-io/scripts/find_aws_icon.py ec2
python ~/.claude/skills/draw-io/scripts/find_aws_icon.py lambda
8. Checklist
- No background color set (page="0")
- Font size appropriate (larger recommended)
- Arrows placed at back layer
- Arrows not overlapping labels (verify in PNG)
- Arrow start/end sufficiently distant from labels (at least 20px)
- Arrows not penetrating boxes or icons (verify in PNG)
- Internal elements not overflowing background frame (verify in PNG)
- 30px+ margin between background frame and internal elements
- AWS service names are official names/correct abbreviations
- AWS icons are latest version (mxgraph.aws4.*)
- No unnecessary elements remaining
- Visually verified PNG conversion
9. Image Display in reveal.js Slides
Add auto-stretch: false to YAML header:
---
title: "Your Presentation"
format:
revealjs:
auto-stretch: false
---
This ensures correct image display on mobile devices.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
openrlhf-training
High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.
gguf-quantization
GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.
Claude Code Guide
Master guide for using Claude Code effectively. Includes configuration templates, prompting strategies "Thinking" keywords, debugging techniques, and best practices for interacting with the agent.
qdrant-vector-search
High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.
behavioral-modes
AI operational modes (brainstorm, implement, debug, review, teach, ship, orchestrate). Use to adapt behavior based on task type.
Didn't find tool you were looking for?