Agent skill
mermaid-export
Export Mermaid diagrams from documents to PNG images. MUST be run immediately after creating or editing any Mermaid diagram, BEFORE committing. Processes .md, .html, .mdx, .rst, .adoc files.
Install this agent skill to your Project
npx add-skill https://github.com/sparkling/claude-config/tree/main/skills/mermaid-export
SKILL.md
Mermaid Diagram Export Skill
Render Mermaid diagrams in documents to PNG images using the mermaid-renderer tool.
CRITICAL RULE: Always render diagrams immediately after creating or editing Mermaid code. Never commit a document with unrendered Mermaid diagrams.
Tool Location
~/.claude/tools/mermaid-renderer/
├── render-mermaid.js # Single diagram renderer
└── process-document.js # Full document processor
Workflow
Step 1: Create Document with Mermaid Diagrams
When creating a document with Mermaid diagrams, use the standard mermaid skill to generate the diagram code:
# My Document
Here's the architecture:
```mermaid
flowchart LR
A[Client] --> B[API Gateway]
B --> C[Service]
More content...
### Step 2: Export Diagrams to PNG
After creating/editing a document with Mermaid blocks, run:
```bash
node ~/.claude/tools/mermaid-renderer/process-document.js <document-path> --verbose
Options:
--format=<fmt>- Output format:png(default) orsvg--theme=<theme>- Mermaid theme:default,forest,dark,neutral--verbose- Show detailed progress--dry-run- Preview without making changes
NOTE: PNG is the default format for maximum compatibility with markdown previewers (VS Code, GitHub, Zed). Use --format=svg only when scalable vector output is specifically needed.
Step 3: Result Structure
The tool will:
- Extract all
\``mermaid` code blocks - Render each to SVG in
diagrams/{document-name}/ - Replace the code block with an image reference
- Preserve original code in a comment for future editing
Output structure:
document.md
diagrams/
└── document/
├── diagram-1.png
├── diagram-2.png
└── architecture-diagram.png
Document transformation:
<!-- Before -->
` ``mermaid
flowchart LR
A --> B
` ``
<!-- After -->

<details>
<summary>Mermaid Source</summary>
` ``mermaid
flowchart LR
A --> B
` ``
</details>
IMPORTANT: Use a collapsible <details> block (not HTML comments) to preserve the source. This keeps the document clean while making the source visible and editable.
Rendering Single Diagrams
To render a single diagram without processing a full document:
# From a .mmd file
node ~/.claude/tools/mermaid-renderer/render-mermaid.js diagram.mmd output.svg
# From stdin
echo "flowchart TD; A-->B" | node ~/.claude/tools/mermaid-renderer/render-mermaid.js --stdin output.svg
# With theme
node ~/.claude/tools/mermaid-renderer/render-mermaid.js diagram.mmd output.svg --theme=dark
Supported Document Types
| Extension | Image Syntax Used |
|---|---|
.md, .markdown |
 with HTML comment |
.html, .htm |
<img> tag with HTML comment |
.mdx |
JSX <img> with JSX comment |
.rst |
.. image:: directive |
.adoc |
image:: macro |
Integration with Mermaid Skill
This skill works together with the main Mermaid diagram skill:
- Design: Use the mermaid skill to design diagrams with proper styling
- Export: Use this skill to render diagrams to PNG
MANDATORY workflow:
- Create/edit mermaid code blocks in document
- Run the export tool immediately (before any commit)
- Commit document AND generated PNGs together
DO NOT commit a document containing unrendered mermaid blocks. The render step is not optional.
Diagram Naming
Diagrams are named based on:
- Title in diagram config (if present)
- First node name in the diagram
- Fallback:
diagram-1,diagram-2, etc.
To control naming, add a title:
---
title: Architecture Overview
---
flowchart LR
...
→ Generates architecture-overview.svg
Editing Rendered Diagrams
To edit a previously rendered diagram:
- Expand the
<details>block to view the mermaid source - Edit the mermaid code directly inside the
<details>block - Re-run the export tool
- The image file is updated in place (document structure unchanged)
The tool automatically detects the rendered pattern ( + <details> with mermaid source) and re-renders the images without modifying the document structure.
ELK Layout Support
This renderer fully supports ELK layout for complex diagrams:
---
config:
layout: elk
elk:
mergeEdges: true
---
flowchart LR
...
How Re-rendering Works
The tool handles two scenarios:
1. Re-rendering Existing Diagrams
When it finds the rendered pattern:

<details>
<summary>Mermaid Source</summary>
` ``mermaid
flowchart LR
A --> B
` ``
</details>
The tool:
- Extracts the mermaid code from inside
<details> - Re-renders the image file in place
- Does NOT modify the document (no nesting issues)
2. Initial Rendering of New Diagrams
When it finds a standalone mermaid block (not inside <details>):
` ``mermaid
flowchart LR
A --> B
` ``
The tool:
- Renders to a new image file
- Wraps the block with image reference +
<details>structure
Troubleshooting
"Could not find Chrome"
Puppeteer needs Chromium. Install globally: npm install -g puppeteer
Diagram doesn't render
- Check Mermaid syntax at https://mermaid.live
- Ensure valid ELK config if using
layout: elk
Diagram renders as 53,854-byte error PNG
This is the mermaid "Syntax error in text" placeholder. Common causes:
classDef class—classis a reserved keyword. Useclsinstead- Angle brackets in comments (
%% <http://...>) — the renderer escapes these automatically, but if rendering standalone, wrap inescapeForHtmlPre() \nin labels — use<br/>instead- Run
bash ~/.claude/tools/mermaid-renderer/validate-diagrams.sh <doc>to test all blocks
SVG too large/small
The tool auto-sizes based on diagram content. For custom sizing, use the <img> tag with width/height attributes (see Image Sizing section below).
HTML/PDF Export (After Rendering)
After rendering diagrams to PNG, export the full document to HTML/PDF:
node ~/.claude/tools/markdown-export/convert.js <document-path> --verbose
The HTML export:
- Embeds images inline as base64 data URIs (self-contained HTML)
- Caps image height at
80vhso diagrams fit on screen without scrolling - Pan/zoom viewer — clicking any image opens an interactive overlay:
- Scroll wheel: zoom centered on cursor
- Click-drag: pan
- Double-click: toggle fit/1:1
- Touch: pinch-zoom and drag
- Keyboard:
+/-zoom,0fit,1actual size,Escclose
- Angle bracket escaping in mermaid
<pre>context (prevents HTML parsing of<url>in comments) - Generates PDF via Puppeteer with rendered diagrams
Image Sizing
CRITICAL: When resizing images in markdown, you MUST use the <img> tag with size attributes. NEVER use CSS.
Correct Way (Use This)
<img src="diagrams/document/diagram.png" alt="Diagram Name" width="90%">
Or with fixed pixel width:
<img src="diagrams/document/diagram.png" alt="Diagram Name" width="600">
Wrong Way (Never Do This)
<!-- WRONG: CSS styling -->
{: style="width: 90%" }
<!-- WRONG: Kramdown attributes -->
{: width="90%" }
Size Guidelines
| Diagram Type | Recommended Width |
|---|---|
| Simple flowcharts | width="70%" |
| Complex architectures | width="90%" |
| Sequence diagrams | width="80%" |
| State machines | width="60%" |
| ER diagrams | width="85%" |
With Details Block
When including the mermaid source in a details block:
<img src="diagrams/document/diagram.png" alt="Architecture Overview" width="90%">
<details>
<summary>Mermaid Source</summary>
` ``mermaid
flowchart LR
A --> B
` ``
</details>
Diagrams not being re-rendered
Ensure the document has the correct pattern:  followed by <details> with <summary>Mermaid Source</summary> and the mermaid code block. The pattern matching is strict to avoid false positives.
Example Commands
# Process a markdown file
node ~/.claude/tools/mermaid-renderer/process-document.js ./docs/README.md --verbose
# Process with dark theme
node ~/.claude/tools/mermaid-renderer/process-document.js ./docs/arch.md --theme=dark
# Dry run to see what would happen
node ~/.claude/tools/mermaid-renderer/process-document.js ./docs/README.md --dry-run
# Render single diagram
node ~/.claude/tools/mermaid-renderer/render-mermaid.js ./diagrams/flow.mmd ./output/flow.svg
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
notebook
Create professional Jupyter notebooks with zero-execution styling using markdown-based CSS and HTML components following Airbnb DLS principles.
qlever
Query, configure, and optimize QLever—the high-performance open-source RDF triplestore. Covers SPARQL queries, GeoSPARQL, text search, CLI operations, and public endpoints. Informed by Kurt Cagle's semantic web expertise.
dot-export
Export DOT/Graphviz diagrams from documents to PNG images (default) or SVG. Use after creating documents with DOT code blocks when you need to render them as static images for distribution, GitHub, wikis, or static sites. Supports multiple layout engines (dot, neato, circo, fdp, twopi, osage, patchwork). Processes .md, .html, .mdx, .rst, .adoc files.
owl
Design and implement OWL 2 ontologies for knowledge representation. Covers RDFS foundations, OWL profiles (EL, QL, RL), class expressions, property characteristics, reasoning patterns, SKOS taxonomies, and the evolving shift toward SHACL. Informed by Kurt Cagle's ontologist perspective.
diagramming
Create professional diagrams using Mermaid or DOT/Graphviz. Mermaid for flowcharts, sequences, classes, ER, Gantt, architecture with semantic coloring and WCAG AA accessibility. DOT/Graphviz for pure network graphs, semantic webs, and maximum layout control.
skos
Design, implement, and manage knowledge organization systems with SKOS (Simple Knowledge Organization System). Covers concept schemes, taxonomies, thesauri, controlled vocabularies, semantic relations, mapping, SKOS-XL labels, AI/LLM integration patterns, and SHACL validation. Informed by Kurt Cagle's ontologist perspective and W3C specifications.
Didn't find tool you were looking for?