一键导入
mind-map
Use when user asks to create, visualize, or update a mind map, architecture diagram, or relationship diagram for any project or topic.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Use when user asks to create, visualize, or update a mind map, architecture diagram, or relationship diagram for any project or topic.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | mind-map |
| description | Use when user asks to create, visualize, or update a mind map, architecture diagram, or relationship diagram for any project or topic. |
| user_invocable | true |
Generate .{name}.monkeymap/ folder (dotfile, name derived from topic) and open in Monkey Map. Examples: .architecture.monkeymap/, .pipeline.monkeymap/. Multiple maps per project are supported.
Nodes are short labels (3-6 words). All substance goes in node detail files (nodes/{id}.md) -- users click a node to read details in a side panel. A 1000-line doc should produce 15-25 nodes with rich details per node.
Read the source. Identify the dominant pattern:
| Pattern | Spine | Edge style |
|---|---|---|
| Process flow (pipeline, workflow) | Linear chain: step->step | straight |
| Hierarchy (tree, taxonomy, features) | Root fans out to children | default (bezier) |
| Dashboard (status, audit, tracking) | Grouped regions side by side | default |
Pipeline steps chain sequentially -- do NOT connect all steps back to a hub.
| Use | Type | Config |
|---|---|---|
| Process step | shape | shape: "box" |
| Section header | mindmap | bold: true |
| Leaf/data point | mindmap | fontSize: "small" |
| Key metric | shape | shape: "ellipse" |
| Decision point | shape | shape: "diamond" |
| Annotation (max 3-4 per map) | note | 1 sentence, use sparingly |
In the manifest, nodes have hasDetails: true instead of inline details. Details live in nodes/{id}.md.
Label: 3-6 words, what the node IS. Details (in .md file): Multi-line markdown. Can include status, context, gaps, links, decisions. No length limit but keep it focused.
Nodes: green #16a34a = done, red #dc2626 = failed/blocked, yellow #ca8a04 = partial, indigo #4f46e5 = structural, purple #9333ea = conceptual, cyan #0891b2 = data, pink #e11d48 = edge cases
Notes: #fef9c3 = context, #fecaca = problem, #dbeafe = insight, #dcfce7 = verified
straight with markerEnd: { type: "arrowclosed", width: 16, height: 16 }default with arrowdefault, NO arrow.topic-name.monkeymap/
manifest.json # app-facing: meta + nodes (no details) + edges
map.md # agent-facing: text tree of mind map structure
nodes/
{uuid}.md # plain markdown details (only for nodes with content)
# Create project folder
mkdir -p .topic-name.monkeymap/nodes
# Write manifest (nodes WITHOUT details, hasDetails flag instead)
cat > .topic-name.monkeymap/manifest.json << 'MANIFEST_EOF'
{
"version": 2,
"meta": {
"title": "Diagram Title",
"created": "2026-03-16T00:00:00.000Z",
"viewport": { "x": 0, "y": 0, "zoom": 1 }
},
"nodes": [
{
"id": "uuid1",
"type": "mindmap",
"position": { "x": 250, "y": 250 },
"data": { "label": "Root Topic", "color": "#4f46e5", "hasDetails": true }
}
],
"edges": []
}
MANIFEST_EOF
# Write node details as individual markdown files
cat > .topic-name.monkeymap/nodes/uuid1.md << 'EOF'
Details content here in plain markdown.
Can include multiple paragraphs, lists, code blocks, etc.
EOF
# Generate map.md (text tree for navigation)
cat > .topic-name.monkeymap/map.md << 'EOF'
# Diagram Title
- Root Topic [uuid1]
- Child Node [uuid2]
~ Associated Node [uuid3]
EOF
For 40+ nodes, generate with node -e programmatically.
Auto-generated tree showing hierarchy and associations:
- items = children (vertical connections)~ items = associations (horizontal connections)[uuid] after label = node ID, matches filename in nodes/map.md to understand structure and find target nodesnodes/{id}.md for detailsmanifest.json (that's for the app)map.md to understand current structurenodes/{id}.md directlymanifest.json, modify, write backmap.mdopen -a "Monkey Map" "$(pwd)/.topic-name.monkeymap" # desktop app (macOS)
monkey-map --file .topic-name.monkeymap # CLI fallback (any OS)
Save reusable node/edge arrangements as flow templates:
# Via CLI API
curl -X POST http://localhost:3141/api/flows \
-H "Content-Type: application/json" \
-d '{"name": "My Template", "nodes": [...], "edges": [...]}'
# List saved flows
curl http://localhost:3141/api/flows
In the desktop app, select nodes and use the Flow Picker to save/load templates.