Read the target canvas JSON. Parse nodes and edges arrays.
Collect existing IDs to prevent collisions.
Generate a new ID: [type]-[slug]-[unix-timestamp].
Calculate position using the auto-positioning algorithm (see canvas orchestrator).
Spacing check: Verify the calculated position has at least 80px horizontal and 60px vertical gap from ALL existing content nodes. If not, shift the position until spacing is clear.
Append the new node to the nodes array (after any groups — z-index ordering).
Validate: Run python3 scripts/canvas_validate.py <path> — check for overlaps and node count.
Write the updated canvas JSON.
Report: "Added [description] to [zone] zone at position ([x], [y])."

Node Operations

add image (`/canvas add image [path or url]`)

Resolve the image:

If URL (starts with http): download with curl -sL [url] -o [media_dir]/[filename]. Derive filename from URL path, or use img-[timestamp].jpg if unclear.
If local path outside canvas dir: cp [path] [media_dir]/
If already canvas-relative: use as-is.

Detect aspect ratio:

python3 -c "from PIL import Image; img=Image.open('[path]'); print(img.width, img.height)"
# or fallback
identify -format '%w %h' [path]

Map to the sizing table in ../canvas/references/canvas-spec.md (7 ratios + PDF + fallback).

Create file node with calculated dimensions. Position using auto-layout.

add text (`/canvas add text [content]`)

{
  "id": "text-[slug]-[timestamp]",
  "type": "text",
  "text": "[content]",
  "x": 0, "y": 0,
  "width": 300, "height": 120
}

For multi-line content, estimate height: count newlines, multiply by 24, add 40px padding. Minimum 120px.

add pdf (`/canvas add pdf [path]`)

Copy to [media_dir]/ if outside canvas dir. Fixed size: width=400, height=520.

add note (`/canvas add note [wiki-page]`)

Search for a file matching the page name (case-insensitive, partial match).
Use "type": "file" with vault-relative path. Not "type": "link" — that's for URLs only.
Default size: width=300, height=100.

add link (`/canvas add link [url]`)

{
  "id": "link-[slug]-[timestamp]",
  "type": "link",
  "url": "[url]",
  "x": 0, "y": 0,
  "width": 400, "height": 120
}

Obsidian fetches Open Graph preview automatically.

add mermaid (`/canvas add mermaid [code]`)

Mermaid renders natively in Obsidian text nodes. Wrap in a fenced code block:

{
  "id": "text-mermaid-[timestamp]",
  "type": "text",
  "text": "```mermaid\n[code]\n```",
  "x": 0, "y": 0,
  "width": 500, "height": 400,
  "color": "5"
}

Wider text nodes work better for Mermaid (min 400px wide, 300px tall).

add svg (`/canvas add svg [description or path]`)

If path: copy SVG to media dir, add as file node. Width=400, height based on viewBox aspect ratio.
If description: delegate to /svg skill to generate, then add the output file.
SVGs render as <img> — no interactivity. Must have viewBox attribute.

add gif (`/canvas add gif [description or path]`)

If path: copy GIF to media dir, add as file node.
If description: delegate to /claude-gif-generate skill, then add the output.
Performance limit: max 3 GIFs per canvas, max 480px width. Warn if exceeded.

add banana (`/canvas add banana [prompt]`)

Check if banana skill is available. If not: "Install the banana skill for AI image generation."
Delegate to banana skill with the prompt.
Wait for image generation.
Add the generated image as a file node (detect aspect ratio, auto-size).
Log the path to .recent-images.txt in the canvas directory.

Zone Operations

zone (`/canvas zone [name] [color]`)

Read canvas JSON.
Find max_y of all existing nodes. Use -80 if canvas is empty (consistent with starter canvas layout where title is at y=-300 and default zone at y=-140):
```
if not canvas_nodes:
    max_y = -80
else:
    max_y = max(n["y"] + n.get("height", 0) for n in canvas_nodes) + 60
```
Create group node:

{
  "id": "zone-[slug]-[timestamp]",
  "type": "group",
  "label": "[name]",
  "x": -400,
  "y": "[max_y]",
  "width": 1000,
  "height": 400,
  "color": "[color or '4']"
}

Valid colors: "1"=red "2"=orange "3"=yellow "4"=green "5"=cyan "6"=purple

Insert the group node BEFORE content nodes in the array (z-index: groups render behind).
Write and report.

Edge Operations

connect (`/canvas connect [from] [to] [label]`)

Read canvas JSON.
Find from and to nodes by ID, label text, or partial match.
If ambiguous, list matches and ask the user to clarify.
Create edge:

{
  "id": "e-[from-slug]-[to-slug]-[timestamp]",
  "fromNode": "[from-id]",
  "toNode": "[to-id]",
  "toEnd": "arrow"
}

Omit fromSide/toSide for auto-routing (better results).
Add optional label if provided.
Write and report.

Banana Import

from banana (`/canvas from banana`)

Check [canvas_dir]/.recent-images.txt for recently logged image paths.

If not found or empty: search for images modified in the last 10 minutes:

find [media_dir] -name "*.png" -o -name "*.jpg" -newer /tmp/ten-min-ago

If still none: show the 5 most recently modified images.
Present list: "Found N recent images. Add to canvas? Which zone?"
On confirmation: add each using the add image logic.

Media Directory

Vault mode: _attachments/images/canvas/
Standalone mode: .canvases/assets/
Create if it doesn't exist.
All file node paths must be relative to the vault/project root.

Node Operations

add image (/canvas add image [path or url])

add text (/canvas add text [content])

add pdf (/canvas add pdf [path])

add note (/canvas add note [wiki-page])

add link (/canvas add link [url])

add mermaid (/canvas add mermaid [code])

add svg (/canvas add svg [description or path])

add gif (/canvas add gif [description or path])

add banana (/canvas add banana [prompt])

canvas-populate

canvas-populate: Add Content to Canvases

General Workflow