mermaid-linter

name	mermaid-linter
description	This skill should be used when the user asks to "validate Mermaid diagrams", "lint diagrams", "check diagram syntax", "mermaid health score", or needs to validate Mermaid diagrams in markdown files for syntax errors, configuration, and rendering quality. Uses mcp-mermaid for full validation and local regex for fast pre-checks.

Mermaid Diagram Linter

Validate Mermaid diagrams in markdown files for syntax errors, text overflow issues, configuration requirements, and health score. Integrates with mcp-mermaid MCP server for full syntax validation and local regex pre-checks for fast CI/pre-commit use.

When to Use

• Before deploying documentation sites
• When diagrams appear as raw code instead of rendered
• To validate diagram text length (prevents overflow)
• Part of documentation quality checks (/folio:docs:check Phase 5)
• Release gating via health score

Validation Checks

1. Configuration Validation (mkdocs.yml)

Check that MkDocs is properly configured for enhanced markdown features.

✅ RECOMMENDED: Native Integration (Material for MkDocs)

Per official Material docs, native Mermaid integration requires ONLY the superfences extension:

# Required in mkdocs.yml - Native integration
markdown_extensions:
  - pymdownx.superfences:
      custom_fences:
        - name: mermaid
          class: mermaid
          format: !!python/name:pymdownx.superfences.fence_code_format
  - pymdownx.emoji:
      emoji_index: !!python/name:material.extensions.emoji.twemoji
      emoji_generator: !!python/name:material.extensions.emoji.to_svg
  - attr_list          # Required for icon attributes { .lg .middle }
  - md_in_html         # Required for card grids and markdown in HTML

# NO extra_javascript needed - superfences handles it natively!

⚠️ WARNING: Adding extra_javascript CDN can cause conflicts

If you have this in your config, remove it:

extra_javascript:
  - https://unpkg.com/mermaid@10/dist/mermaid.min.js  # ❌ NOT NEEDED

Material for MkDocs handles Mermaid natively. Adding the CDN manually can cause:

• Double initialization
• Version conflicts
• Theme styling issues

Only add extra_javascript if:

• Using custom Mermaid configuration (ELK layouts, etc.)
• Need specific Mermaid version for compatibility
• Using non-Material theme

If custom config is needed, use:

extra_javascript:
  - javascripts/mermaid-config.js  # Your custom config

With javascripts/mermaid-config.js:

import mermaid from 'https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.esm.min.mjs';
mermaid.initialize({ startOnLoad: false, securityLevel: 'loose' });

Common Configuration Issues:

Missing Extension	Symptom	Fix
custom_fences	Mermaid renders as code blocks	Add superfences custom_fences
pymdownx.emoji	:rocket: shows as text	Add emoji extension
attr_list	Icon attributes not working	Add attr_list
md_in_html	Card grids show as raw text	Add md_in_html

Material for MkDocs Card Grids:

Card grids require BOTH attr_list AND md_in_html:

<div class="grid cards" markdown>

-   :material-rocket:{ .lg .middle } **Title**

    Description text

</div>

Without md_in_html, the markdown inside <div> tags won't be processed!

2. CSS Validation

Check for Mermaid overflow CSS in stylesheets:

/* Required for proper rendering */
.mermaid {
  overflow-x: auto;
  text-align: center;
}

.mermaid svg {
  max-width: 100%;
  height: auto;
}

.mermaid .nodeLabel {
  font-size: 12px !important;
}

3. Syntax Validation

Per official Mermaid docs, check each diagram for valid syntax:

Diagram Types:

• Fully supported by Material theme: flowchart, sequenceDiagram, classDiagram, stateDiagram, erDiagram
• Work but no theme styling: graph, gantt, pie, journey, gitGraph, mindmap, quadrantChart

Direction Keywords:

• TB or TD - Top to bottom (recommended for mobile)
• BT - Bottom to top
• LR - Left to right (simple linear flows only)
• RL - Right to left

Critical Syntax Rules:

1. Reserved "end" keyword - Must capitalize to avoid breaking diagram:

   flowchart TD
       A[Start] --> B[End]      ✅ Capitalized "End"
       A[Start] --> B[end]      ❌ BREAKS diagram
   

2. First letter 'o' or 'x' - Add space or capitalize:

   flowchart LR
       A --> oNode    ❌ Treated as circle edge
       A --> O_Node   ✅ Capitalize or use underscore
   

3. Prefer flowchart over graph - Both work identically, but flowchart is clearer intent

4. All nodes must be connected - Orphaned nodes cause syntax errors:

   flowchart TD
       A --> B
       C[Orphan]      ❌ Not connected to anything
   

4. Text Formatting Best Practices

Per official Mermaid docs, use automatic text wrapping instead of manual line breaks.

❌ AVOID: Manual <br/> tags

flowchart TD
    A[Getting Started<br/>7 steps]    ❌ Manual breaks
    B[Intermediate<br/>11 steps]

✅ RECOMMENDED: Markdown strings with automatic wrapping

flowchart TD
    A["`**Getting Started**
    7 steps · 10 minutes
    Essential commands`"]           ✅ Auto-wraps, supports markdown

Markdown string syntax (backticks with quotes):

• Supports bold, italic, code
• Automatic text wrapping at node width
• More maintainable than <br/> tags
• Better mobile responsiveness

Text Length Guidelines:

Element	Max Length	Recommendation
Node text (single line)	15-20 chars	Use abbreviations, ... for paths
Edge labels	10 chars	Keep brief
Subgraph titles	20 chars	Short descriptive names

Examples:

%% BAD: Text too long, will overflow
graph LR
    A["~/.git-worktrees/aiterm/feature-mcp/"] --> B

%% GOOD: Abbreviated with markdown string
flowchart LR
    A["`**Worktrees**
    ~/.git-worktrees/
    .../feature-mcp`"] --> B["`Next Step`"]

5. Layout Direction Best Practices

Prefer vertical layouts (TD/TB) over horizontal (LR/RL) for complex diagrams:

Layout	Best For	Avoid For
flowchart TD	Complex workflows, decision trees, multi-branch flows	-
flowchart LR	Simple linear flows (< 5 nodes), system architecture	Complex workflows, parallel processes

Why vertical layouts render better:

• Better spacing on mobile/narrow screens
• Reduced text overlap in decision diamonds
• Clearer visualization of parallel processes using subgraphs
• More readable for ADHD-friendly documentation

Examples:

%% BAD: Horizontal complex diagram
flowchart LR
    A --> B --> C --> D
    B --> E
    C --> F
    D --> G
    E --> G
    F --> G

%% GOOD: Vertical complex diagram
flowchart TD
    A --> B --> C --> D
    B --> E
    C --> F
    D --> G
    E --> G
    F --> G

When to use subgraphs (always with TD):

• Parallel worktree workflows
• Multi-path decision flows
• Feature development pipelines

6. Diagram Count Report

Report all mermaid diagrams found:

• Total count per file
• Diagram types used
• Files with most diagrams
• Diagrams using <br/> tags (flag for modernization)

7. MCP Validation (via mcp-mermaid)

When the mcp-mermaid MCP server is available, use it for full syntax validation beyond regex pre-checks:

• Syntax validation — catches errors regex cannot detect
• Rendering test — confirms diagram renders to SVG
• Error reporting — detailed error messages with fix suggestions

The MCP server is configured in .mcp.json:

{
  "mcpServers": {
    "mcp-mermaid": {
      "command": "npx",
      "args": ["-y", "mcp-mermaid"]
    }
  }
}

8. Health Score

Composite quality metric (0-100) for release gating:

health_score = syntax_validity*0.5 + best_practices*0.3 + rendering_success*0.2

Score	Level	Release Gate
>= 90	Good	Pass
>= 80	Warning	Pass (default threshold)
< 80	Fail	Blocked

Validation Commands

# Automated validation (recommended)
python3 scripts/mermaid-validate.py docs/ commands/ skills/

# With health score
python3 scripts/mermaid-validate.py docs/ --health-score

# Release gate check
python3 scripts/mermaid-validate.py docs/ --gate

# Errors only (fast, for CI)
python3 scripts/mermaid-validate.py docs/ --errors-only

# Auto-fix safe patterns (dry-run)
python3 scripts/mermaid-autofix.py docs/

# Auto-fix (apply changes)
python3 scripts/mermaid-autofix.py docs/ --fix

# Manual grep checks
grep -r "^\`\`\`mermaid" docs/ | wc -l

# Find diagrams using deprecated 'graph' keyword
grep -A1 "^\`\`\`mermaid" docs/**/*.md | grep -c "^graph "

Output Format

MERMAID DIAGRAM VALIDATION
==========================

Configuration:
  ✅ custom_fences configured in mkdocs.yml
  ✅ mermaid.js CDN in extra_javascript
  ⚠️  No Mermaid CSS found in stylesheets

Diagrams Found: 15 total
  docs/guides/GIT-WORKTREES-GUIDE.md: 12 diagrams
  docs/architecture/AITERM-ARCHITECTURE.md: 3 diagrams

Syntax Validation:
  ✅ All 15 diagrams have valid syntax

Text Length Issues:
  ⚠️  docs/guides/GIT-WORKTREES-GUIDE.md:156
      Node text "~/.git-worktrees/aiterm/feature-mcp/" is 38 chars (max 15)
      Suggestion: Use "~/.git-worktrees/.../feature-mcp"

Health Score: 92/100 (Good)
  Syntax validity:    100.0%
  Best practices:     80.0%
  Rendering success:  100.0%

Summary:
  Passed: 14/15
  Warnings: 1 (text length)
  Errors: 0 (syntax)
  Config: 2/3 (missing CSS)
  Release gate: PASS (>= 80)

Recommendations:
  1. Add Mermaid CSS to docs/stylesheets/extra.css
  2. Shorten node text in GIT-WORKTREES-GUIDE.md line 156

Integration

Called by:

• /folio:docs:check - Phase 5: Mermaid Validation
• /craft:site:status --check - Site health check
• /folio:docs:validate - Documentation validation
• /craft:check commit - Pre-commit checks
• /craft:site:deploy - Health score gate (>= 80)

Related:

• /folio:docs:mermaid - Diagram templates, NL creation, MCP validation
• /craft:site:status - Quick Mermaid config check
• /folio:docs:generate - Includes Mermaid guidelines
• mermaid-expert agent - MCP-powered diagram generation

Scripts:

• scripts/mermaid-validate.py - Extraction + regex pre-checks + health score
• scripts/mermaid-autofix.py - Safe auto-fix engine

MCP Server:

• mcp-mermaid - Full syntax validation and SVG/PNG rendering

Best Practices

Per official Mermaid documentation and Material for MkDocs:

1. Use native integration - Let Material handle Mermaid, don't add extra_javascript unless needed
2. Markdown strings over <br/> - Use backtick syntax for multi-line text with auto-wrapping
3. Prefer flowchart keyword - Clearer intent than graph (both work identically)
4. Vertical layouts (TD) - Better mobile rendering, reduced text overlap
5. Capitalize "end" keyword - Use "End" or "END" to avoid breaking diagrams
6. Connect all nodes - Orphaned nodes cause syntax errors
7. Test locally - Run mkdocs serve to preview diagrams before deployment
8. Use subgraphs - Group related nodes for clarity and structure
9. Consistent diagram types - Stick to one type per document (flowchart, sequence, etc.)
10. Add comments - Use %% for documentation within diagrams
11. Keep it simple - Avoid over-engineering, Swedish "lagom" principle applies

Common Issues

Issue	Cause	Fix
Diagrams show as code	Missing custom_fences	Add superfences to mkdocs.yml
Diagram broken ("end" keyword)	Used lowercase "end"	Capitalize to "End" or "END"
Orphaned node error	Node not connected	Connect with arrow or remove
Text overflows boxes	Using <br/> tags	Switch to markdown strings with backticks
Double rendering	Extra_javascript + native	Remove extra_javascript, use native integration
Poor mobile rendering	Horizontal (LR) layout	Switch to vertical (TD) layout
Theme colors not applied	Using unsupported diagram type	Use flowchart/sequence/class/state/ER only
Syntax error on 'o' or 'x'	First letter treated as edge	Capitalize or add space before name