Index of every page on this site: llms.txt
All skills
axtonliu avatar

/obsidian-canvas-creator

@1265976
Run it yourself

Create Obsidian Canvas files from text content, supporting both MindMap and freeform layouts. Use this skill when users want to visualize content as an interactive canvas, create mind maps, or organize information spatially in Obsidian format.

Use this Skill: https://skilld.dev/gh/axtonliu/axton-obsidian-visual-skills/obsidian-canvas-creator

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈67 tokens for metadata: the name and description. ≈1.6k when used: this file. ≈6.4k more on demand in 2 files.

Description uses 3.4% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈269 characters, or 3.4%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Obsidian Canvas Creator

Transform text content into structured Obsidian Canvas files with support for MindMap and freeform layouts.

When to Use This Skill

  • User requests to create a canvas, mind map, or visual diagram from text
  • User wants to organize information spatially
  • User mentions "Obsidian Canvas" or similar visualization tools
  • Converting structured content (articles, notes, outlines) into visual format

Core Workflow

1. Analyze Content

Read and understand the input content:

  • Identify main topics and hierarchical relationships
  • Extract key points, facts, and supporting details
  • Note any existing structure (headings, lists, sections)

2. Determine Layout Type

Ask user to choose or infer from context:

MindMap Layout:

  • Radial structure from center
  • Parent-child relationships
  • Clear hierarchy
  • Good for: brainstorming, topic exploration, hierarchical content

Freeform Layout:

  • Custom positioning
  • Flexible relationships
  • Multiple connection types
  • Good for: complex networks, non-hierarchical content, custom arrangements

3. Plan Structure

For MindMap:

  • Identify central concept (root node)
  • Map primary branches (main topics)
  • Organize secondary branches (subtopics)
  • Position leaf nodes (details)

For Freeform:

  • Group related concepts
  • Identify connection patterns
  • Plan spatial zones
  • Consider visual flow

4. Generate Canvas

Create JSON following the Canvas specification:

Node Creation:

  • Assign unique 8-12 character hex IDs
  • Set appropriate dimensions based on content length
  • Apply consistent color schemes
  • Ensure no coordinate overlaps

Edge Creation:

  • Connect parent-child relationships
  • Use appropriate arrow styles
  • Add labels for complex relationships
  • Choose line styles (straight for hierarchy, curved for cross-references)

Grouping (Optional):

  • Create visual containers for related nodes
  • Use subtle background colors
  • Add descriptive labels

5. Apply Layout Algorithm

MindMap Layout Calculations:

Refer to references/layout-algorithms.md for detailed algorithms. Key principles:

  • Center root at (0, 0)
  • Distribute primary nodes radially
  • Space secondary nodes based on sibling count
  • Maintain minimum spacing: 320px horizontal, 200px vertical

Freeform Layout Principles:

  • Start with logical groupings
  • Position groups with clear separation
  • Connect across groups with curved edges
  • Balance visual weight across canvas

6. Validate and Output

Before outputting:

Validation Checklist:

  • All nodes have unique IDs
  • No coordinate overlaps (check distance > node dimensions + spacing)
  • All edges reference valid node IDs
  • Groups (if any) have labels
  • Colors use consistent format (hex or preset numbers)
  • JSON is properly escaped (Chinese quotes: 『』 for double, 「」 for single)

Output Format:

  • Complete, valid JSON Canvas file
  • No additional explanation text
  • Directly importable into Obsidian

Node Sizing Guidelines

Text Length-Based Sizing:

  • Short text (<30 chars): 220 × 100 px
  • Medium text (30-60 chars): 260 × 120 px
  • Long text (60-100 chars): 320 × 140 px
  • Very long text (>100 chars): 320 × 180 px

Color Schemes

Preset Colors (Recommended):

  • "1" - Red (warnings, important)
  • "2" - Orange (action items)
  • "3" - Yellow (questions, notes)
  • "4" - Green (positive, completed)
  • "5" - Cyan (information, details)
  • "6" - Purple (concepts, abstract)

Custom Hex Colors: Use for brand consistency or specific themes. Always use uppercase format: "#4A90E2"

Critical Rules

  1. Quote Handling:

    • Chinese double quotes → 『』
    • Chinese single quotes → 「」
    • English double quotes → \"
  2. ID Generation:

    • 8-12 character random hex strings
    • Must be unique across all nodes and edges
  3. Z-Index Order:

    • Output groups first (bottom layer)
    • Then subgroups
    • Finally text/link nodes (top layer)
  4. Spacing Requirements:

    • Minimum horizontal: 320px between node centers
    • Minimum vertical: 200px between node centers
    • Account for node dimensions when calculating
  5. JSON Structure:

    • Top level contains only nodes and edges arrays
    • No extra wrapping objects
    • No comments in output
  6. No Emoji:

    • Do not use any Emoji symbols in node text
    • Use color coding or text labels for visual distinction instead

Examples

Simple MindMap Request

User: "Create a mind map about solar system planets"

Process:

  1. Identify center: "Solar System"
  2. Primary branches: Inner Planets, Outer Planets, Dwarf Planets
  3. Secondary nodes: Individual planets with key facts
  4. Apply radial layout
  5. Generate JSON with proper spacing

Freeform Content Request

User: "Turn this article into a canvas" + [article text]

Process:

  1. Extract article structure (intro, body sections, conclusion)
  2. Identify key concepts and relationships
  3. Group related sections spatially
  4. Connect with labeled edges
  5. Apply freeform layout with clear zones

Reference Documents

  • Canvas Specification: references/canvas-spec.md - Complete JSON Canvas format specification
  • Layout Algorithms: references/layout-algorithms.md - Detailed positioning algorithms for both layout types

Load these references when:

  • Need specification details for edge cases
  • Implementing complex layout calculations
  • Troubleshooting validation errors

Tips for Quality Canvases

  1. Keep text concise: Each node should be scannable (<2 lines preferred)
  2. Use hierarchy: Group by importance and relationship
  3. Balance the canvas: Distribute nodes to avoid clustering
  4. Strategic colors: Use colors to encode meaning, not just decoration
  5. Meaningful connections: Only add edges that clarify relationships
  6. Test in Obsidian: Verify the output opens correctly

Common Pitfalls to Avoid

  • Overlapping nodes (always check distances)
  • Inconsistent quote escaping (breaks JSON parsing)
  • Missing group labels (causes sidebar navigation issues)
  • Too much text in nodes (use file nodes for long content)
  • Duplicate IDs (each must be unique)
  • Unconnected nodes (unless intentional islands)

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts1mo5 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    The skill is designed to convert text into Obsidian Canvas files. The only identified concern is the inherent risk of processing untrusted text, which could potentially contain instructions intended to manipulate the agent's behavior during the conversion process.

  • Socket1mo

    No alerts

  • Snyk1mo

    Risk: LOW · No issues

  • Runlayer7mo

    1/5 files flagged

  • ZeroLeaks6mo

    1 finding · Score: 86/100

Signed by skilld at 1265976. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last week.

Steadyupdated 9 months ago

README badge

README badge for axtonliu/axton-obsidian-visual-skills/obsidian-canvas-creator