---

name: mermaid-diagrams description: Create Mermaid diagrams. ONLY use when user explicitly says "Mermaid". NOT for general diagrams, schemas, ASCII art, or wireframes.

Mermaid Diagrams Skill

This skill helps create clean, well-organized Mermaid diagrams for software engineering and architecture visualization.

<when_to_use> Use this skill specifically for creating, editing, fixing, or improving Mermaid diagrams:

• Create new Mermaid diagrams
• Edit existing Mermaid diagrams
• Fix broken or incorrect diagram syntax
• Improve diagram organization or readability
• Visualize software architecture or technical concepts

For simply reading or interpreting existing diagrams, proceed directly without this skill. </when_to_use>

<diagram_types>

This skill supports all major Mermaid diagram types for software engineering:

1. Flowcharts - Process flows, algorithms, decision trees
2. Sequence Diagrams - API interactions, service communication, workflows
3. Class Diagrams - Object models, domain design, relationships
4. ER Diagrams - Database schemas, data models
5. State Diagrams - State machines, workflow states, lifecycle
6. C4 Diagrams - Software architecture (context, container, component)
7. Git Graphs - Branching strategies, version control workflows </diagram_types>

<references> **Parallel Loading:** When creating diagrams, read `references/gotchas.md` AND the relevant syntax reference file simultaneously. Load multiple template files in parallel when creating complex diagrams.

Quick Reference Priority

When creating or editing diagrams, consult references in this order:

For all diagrams first:

1. references/gotchas.md - Common errors, special characters, reserved keywords
2. references/styling.md - Themes, colors, formatting

Then consult the specific syntax reference:

• references/syntax-flowchart.md
• references/syntax-sequence.md
• references/syntax-class.md
• references/syntax-er.md
• references/syntax-state.md
• references/syntax-c4.md
• references/syntax-git.md

For architectural patterns:

• references/patterns.md - Common software architecture patterns

When to Load References

Always load gotchas.md when:

• Creating complex diagrams
• Fixing syntax errors
• Dealing with special characters or reserved words

Load styling.md when:

• User requests specific colors or themes
• Creating presentation-quality diagrams
• Need to emphasize specific elements

Load specific syntax reference when:

• Creating a diagram type for the first time
• User asks about specific features
• Need to verify correct syntax

Load patterns.md when:

• Visualizing software architecture
• User mentions specific patterns (microservices, hexagonal, CQRS, etc.)
• Creating system design diagrams </references>

<templates>

Pre-built, well-commented templates are available in assets/templates/:

• flowchart-template.md - Organized flowchart with sections
• sequence-template.md - Sequence diagram with best practices
• class-template.md - Domain model class diagram
• er-template.md - Database schema ER diagram

Use templates when:

• Starting a new complex diagram
• User wants a well-organized structure
• Creating diagrams that will grow over time </templates>

<best_practices>

1. Clean Code Organization

Use the same principles as code:

• Comment liberally - Use %% to explain sections
• Group related items - Keep related nodes/classes together
• Separate concerns - Define structure first, styling last
• Use descriptive names - Make IDs and labels meaningful

Example:

flowchart TB
    %% =================================================================
    %% USER AUTHENTICATION FLOW
    %% =================================================================
    
    %% -----------------------------------------------------------------
    %% Entry Point
    %% -----------------------------------------------------------------
    Start[User enters credentials]
    
    %% -----------------------------------------------------------------
    %% Validation
    %% -----------------------------------------------------------------
    Validate[Validate input format]
    CheckDB{Credentials valid?}
    
    %% ... rest of diagram

2. Make Diagrams Navigable

For large diagrams:

• Use clear section headers in comments
• Group logically related elements
• Keep consistent indentation in the code
• Add blank lines between sections

Structure pattern:

%% Section 1: Definition
[define elements]

%% Section 2: Connections
[define relationships]

%% Section 3: Styling
[define styles]

3. Handle Complexity

When diagrams get large:

• Break into multiple smaller diagrams
• Use subgraphs for logical grouping
• Link between diagrams with notes
• Consider creating an overview diagram

4. Avoid Common Pitfalls

Before finalizing any diagram, check:

• No reserved keywords as node IDs
• Special characters properly escaped or quoted
• All brackets/parens balanced
• All blocks (subgraph, loop, alt) properly closed
• Unique IDs for all elements
• Consistent quote style throughout </best_practices>

<workflow>

Step 1: Understand Requirements

• What concept needs visualization?
• Which diagram type is most appropriate?
• What's the target audience?

Step 2: Choose Diagram Type

• Process/Algorithm → Flowchart
• Time-based interactions → Sequence diagram
• Object relationships → Class diagram
• Data model → ER diagram
• State management → State diagram
• System architecture → C4 diagrams
• Git workflow → Git graph

Step 3: Start Simple

Begin with basic structure:

flowchart LR
    A --> B
    B --> C

Step 4: Add Details Incrementally

• Add more nodes/relationships
• Add labels and descriptions
• Group into subgraphs if needed
• Add comments for clarity

Step 5: Apply Styling (Optional)

• Choose appropriate theme
• Add colors for semantic meaning
• Highlight important elements
• Ensure accessibility

Step 6: Review and Refine

• Test the diagram renders correctly
• Check for syntax errors
• Verify it communicates the intended message
• Add comments for future maintainability </workflow>

<syntax_reference>

Escaping Special Characters

%% Use quotes for special characters
A["Function with (parentheses)"]
B["Text with [brackets]"]

%% Or use HTML entities
C[getData#40;#41;]  %% getData()

Common Relationship Patterns

Flowcharts:

• --> solid arrow
• -.-> dotted arrow
• ==> thick arrow

Class diagrams:

• <|-- inheritance
• *-- composition
• o-- aggregation
• --> association
• ..> dependency

ER diagrams:

• ||--|| one-to-one
• ||--o{ one-to-many
• }o--o{ many-to-many </syntax_reference>

<use_cases>

API Documentation

Use sequence diagrams to show request/response flows:

sequenceDiagram
    Client->>API: POST /users
    API->>Database: INSERT user
    Database-->>API: User created
    API-->>Client: 201 Created

Database Schema

Use ER diagrams with full attribute details:

erDiagram
    USER {
        uuid id PK
        string email UK
    }
    ORDER {
        uuid id PK
        uuid user_id FK
    }
    USER ||--o{ ORDER : places

System Architecture

Use C4 diagrams for different abstraction levels:

C4Context
    Person(user, "User")
    System(sys, "System")
    Rel(user, sys, "Uses")

Process Documentation

Use flowcharts with clear decision points:

flowchart TB
    Start --> Check{Valid?}
    Check -->|Yes| Process
    Check -->|No| Error

</use_cases>

<troubleshooting>

Diagram Won't Render

1. Check for unbalanced quotes or brackets
2. Look for reserved keywords as IDs
3. Verify all blocks are closed (end statements)
4. Check for special characters - escape them
5. Ensure proper syntax for diagram type

Common Error Fixes

"Parse error" → Check syntax matches diagram type

"Syntax error in text" → Escape special characters or use quotes

Blank output → Check browser console, usually invalid syntax

Wrong appearance → Verify correct diagram type declaration </troubleshooting>

<editing_tips>

When editing existing diagrams:

1. Preserve the original structure and organization
2. Keep existing comments - they're valuable context
3. Maintain consistent naming conventions
4. Don't break working syntax to "improve" it
5. Test after each major change

When creating new diagrams:

1. Start with a template if appropriate
2. Build incrementally - test as you go
3. Add comments explaining the purpose
4. Use semantic styling (not just decoration)
5. Consider maintainability </editing_tips>

<integration>

Mermaid diagrams work great in:

• Documentation - README.md, docs/ folder
• PRs/Issues - Visualize changes or problems
• ADRs - Architecture Decision Records
• Wikis - Team knowledge base
• Presentations - Export as images

They can be version controlled, reviewed, and updated like code! </integration>

<summary>

• Start simple - Add complexity incrementally
• Comment well - Future you will thank you
• Test often - Verify syntax as you build
• Use templates - Don't reinvent the wheel
• Consult gotchas - Avoid common errors
• Stay organized - Group and section your diagrams

The goal is to create diagrams that are:

• Clear - Communicate effectively
• Maintainable - Easy to update later
• Navigable - Easy to understand the code
• Correct - Render without errors

</summary>