excalidraw（Excalidraw 图表绘制）

Excalidraw Diagram Generator (MCP Edition)

Create diagrams on a live Excalidraw canvas using MCP tools. The canvas runs in a browser and updates in real time.

Mental Model

You are placing shapes on a 2D canvas and drawing arrows between them.

(0,0) ────────── x increases ──────────►
  │
  │   ┌──────────┐      ┌──────────┐
  │   │  Box A   │─────►│  Box B   │
  │   └──────────┘      └──────────┘
  │         │
  y         ▼
  increases ┌──────────┐
  │         │  Box C   │
  ▼         └──────────┘

Everything is (x, y, width, height). That's it.

The 5 Tools You Actually Need

Other useful tools: describe_scene (text description of canvas), create_from_mermaid (quick diagram from Mermaid syntax), export_scene (save as.excalidraw JSON file), set_viewport (zoom/pan to fit), export_to_excalidraw_url (shareable link).

How Shapes Work

A shape has: type, position (x, y), size (width, height), colors, and label text.

{
  "type": "rectangle",
  "id": "my-box",
  "x": 100,
  "y": 100,
  "width": 180,
  "height": 70,
  "backgroundColor": "#a5d8ff",
  "strokeColor": "#1971c2",
  "roughness": 0,
  "text": "My Service\nPort 8080"
}

Key points:

• text puts a label directly inside the shape (MCP handles the binding for you)
• roughness: 0 = clean lines. roughness: 1 = hand-drawn look.
• Use \n for multi-line labels
• Shapes: rectangle, ellipse, diamond, text (standalone)

How Arrows Work

Arrows connect shapes by ID. The MCP server auto-routes them to shape edges.

{
  "type": "arrow",
  "x": 0,
  "y": 0,
  "startElementId": "my-box",
  "endElementId": "other-box",
  "strokeColor": "#1971c2",
  "text": "HTTP"
}

Key points:

• startElementId / endElementId = the id of the shape to connect to
• The arrow auto-routes to the nearest edges. You do NOT calculate edge points.
• x, y are still required but can be approximate — binding overrides them
• text adds a label on the arrow
• strokeStyle: "dashed" = async/optional flows. "dotted" = weak dependency.
• startArrowhead / endArrowhead = "arrow", "dot", "triangle", "bar", or null

Step-by-Step Workflow

Step 1: Understand What to Draw

Read the codebase. Identify:

• Components (services, databases, APIs, queues, frontends)
• Connections (which components talk to which, and how)
• Layers (group related components into rows or zones)

When a sample diagram is provided (ASCII art, text mockup, screenshot, etc.):

• Preserve ALL text and detail from the sample by default. Do not simplify, summarize, or omit labels, annotations, bullet points, or sublabels present in the sample.
• Extract every node's full text (titles, subtitles, tool names, metrics, details) and reproduce it verbatim in shape labels using \n for multiline.
• Preserve section headers, status annotations (e.g. "Fail = Stop & Notify"), and arrow labels exactly as written.
• Size boxes large enough to fit the full text (increase height/width beyond defaults as needed).
• The sample is the source of truth for content — you may improve layout, colors, and styling, but never drop information.

Step 2: Read the Design Guide

mcp__excalidraw__read_diagram_guide()

This returns the color palette, sizing rules, and layout best practices. Use it.

Step 3: Clear and VERIFY the Canvas

mcp__excalidraw__clear_canvas()
mcp__excalidraw__get_canvas_screenshot()  // MUST verify empty!

Critical: Previous diagrams can leave ghost elements. Always screenshot after clearing to confirm the canvas is truly empty before creating new elements. If elements remain, clear again.

Step 4: Plan Your Layout on Paper

Before calling any create tool, sketch the layout mentally:

Vertical flow (most common):
  Row 1 (y=0):    Zone backgrounds (large dashed rectangles)
  Row 2 (y=60):   Entry points / Users
  Row 3 (y=350):  Middle layer (APIs, services)
  Row 4 (y=650):  Data layer (databases, storage)

  Columns: x = 40, 440, 840  (spaced 400px apart for labeled arrows)
  Box size: 230 x 160 (standard)  |  200 x 120 (for decision diamonds)
  Spacing between rows: ~200px gap after accounting for box height
  Spacing between boxes in a row: 180px gap (for arrow labels)

Step 5: Create Everything in One Batch

Call batch_create_elements with ALL elements at once. This ensures arrow bindings resolve correctly (arrows can reference shape IDs created in the same batch).

Order of elements in the array:

1. Zone backgrounds (large dashed rectangles) — so they render behind everything
2. Shapes (rectangles, ellipses, diamonds) — with id set for arrow references
3. Arrows — using startElementId / endElementId
4. Standalone text labels (titles, annotations)

Step 6: Self-Critique Loop (automatic)

Automatically validate and fix layout before presenting to the user. Do NOT show the diagram until this loop passes or 2 rounds complete.

Skip for diagrams with fewer than 6 elements.

6a. Snapshot for rollback

mcp__excalidraw__snapshot_scene()

If fixes make things worse, restore with mcp__excalidraw__restore_snapshot().

6b. Geometric validation (via query_elements)

mcp__excalidraw__query_elements({ type: "all" })

Check element positions and sizes programmatically:

6c. Visual validation (via screenshot)

mcp__excalidraw__get_canvas_screenshot()

Check for issues requiring visual judgment:

6d. Fix and re-check

Fix issues with update_element or delete_element + create_element, then screenshot again. If the fix made things worse, restore the snapshot.

Constraints:

• Max 2 rounds. After 2, present what you have.
• Only fix layout/visual issues. Do NOT change architectural content.
• Log what you fixed AND what remains unfixed.

Step 7: Present to User

mcp__excalidraw__set_viewport({ scrollToContent: true })
mcp__excalidraw__get_canvas_screenshot()

Show the final screenshot and summarize:

• Components and connections generated
• Auto-fixes applied during self-critique (if any)
• Remaining issues (if any)
• Available next actions (export, refine, drill-down)

Step 8: Export (if requested)

mcp__excalidraw__export_to_image({ format: "png", filePath: "/path/to/output.png" })
mcp__excalidraw__export_scene({ filePath: "/path/to/output.excalidraw" })
mcp__excalidraw__export_to_excalidraw_url()  // shareable link

Complete Example: 3-Layer Architecture

This shows exactly what to pass to batch_create_elements:

{
  "elements": [
    // --- ZONE BACKGROUNDS (render behind everything) ---
    {
      "type": "rectangle", "id": "zone-frontend",
      "x": 0, "y": 0, "width": 500, "height": 160,
      "backgroundColor": "#e9ecef", "strokeColor": "#868e96",
      "strokeStyle": "dashed", "opacity": 40, "roughness": 0
    },
    {
      "type": "text", "x": 10, "y": 10,
      "text": "Frontend Layer", "fontSize": 14, "strokeColor": "#868e96"
    },
    {
      "type": "rectangle", "id": "zone-backend",
      "x": 0, "y": 200, "width": 500, "height": 160,
      "backgroundColor": "#eebefa", "strokeColor": "#9c36b5",
      "strokeStyle": "dashed", "opacity": 30, "roughness": 0
    },
    {
      "type": "text", "x": 10, "y": 210,
      "text": "Backend Layer", "fontSize": 14, "strokeColor": "#9c36b5"
    },

    // --- SHAPES (give each an id so arrows can reference them) ---
    {
      "type": "rectangle", "id": "react-app",
      "x": 40, "y": 50, "width": 180, "height": 70,
      "backgroundColor": "#a5d8ff", "strokeColor": "#1971c2", "roughness": 0,
      "text": "React App\nFrontend"
    },
    {
      "type": "rectangle", "id": "api-server",
      "x": 40, "y": 250, "width": 180, "height": 70,
      "backgroundColor": "#d0bfff", "strokeColor": "#7048e8", "roughness": 0,
      "text": "API Server\nExpress.js"
    },
    {
      "type": "rectangle", "id": "database",
      "x": 280, "y": 250, "width": 180, "height": 70,
      "backgroundColor": "#b2f2bb", "strokeColor": "#2f9e44", "roughness": 0,
      "text": "PostgreSQL\nDatabase"
    },

    // --- ARROWS (connect shapes by ID) ---
    {
      "type": "arrow", "x": 130, "y": 120,
      "startElementId": "react-app", "endElementId": "api-server",
      "strokeColor": "#1971c2", "text": "REST API"
    },
    {
      "type": "arrow", "x": 220, "y": 285,
      "startElementId": "api-server", "endElementId": "database",
      "strokeColor": "#2f9e44", "text": "SQL"
    },

    // --- TITLE ---
    {
      "type": "text", "x": 100, "y": -40,
      "text": "System Architecture", "fontSize": 24, "strokeColor": "#1e1e1e"
    }
  ]
}

Complete Example: Data Flow Diagram (Parameter Threading)

Shows a parameter traced through 5 layers with split/converge paths, decision node, and side annotations:

{
  "elements": [
    // --- TITLE ---
    {"type": "text", "x": 20, "y": 10, "text": "Data Flow: parameter_name Threading", "fontSize": 24, "strokeColor": "#1e1e1e"},
    {"type": "text", "x": 20, "y": 48, "text": "Subtitle describing the trace", "fontSize": 16, "strokeColor": "#868e96"},

    // --- WHY SECTION (top-right, first-principles context) ---
    {"type": "rectangle", "id": "why-bg", "x": 460, "y": 80, "width": 440, "height": 310,
     "backgroundColor": "#e9ecef", "strokeColor": "#868e96", "roughness": 0},
    {"type": "text", "x": 480, "y": 95, "text": "WHY: The Problem", "fontSize": 20, "strokeColor": "#e03131"},
    {"type": "text", "x": 480, "y": 135, "text": "1. What currently happens", "fontSize": 16, "strokeColor": "#1e1e1e"},
    {"type": "text", "x": 480, "y": 195, "text": "2. Why it's expensive/wrong", "fontSize": 16, "strokeColor": "#e03131"},
    {"type": "text", "x": 480, "y": 275, "text": "3. Gap in current design", "fontSize": 16, "strokeColor": "#1e1e1e"},
    {"type": "text", "x": 480, "y": 335, "text": "Solution: what this change does", "fontSize": 16, "strokeColor": "#2f9e44"},

    // --- FLOW BOXES (center column, 150px vertical pitch) ---
    {"type": "rectangle", "id": "l1", "x": 60, "y": 420, "width": 300, "height": 65,
     "backgroundColor": "#a5d8ff", "strokeColor": "#1971c2", "roughness": 0,
     "text": "Entry Point\nfile/path.py"},

    // Split into two paths
    {"type": "rectangle", "id": "l2a", "x": -100, "y": 570, "width": 290, "height": 65,
     "backgroundColor": "#a5d8ff", "strokeColor": "#1971c2", "roughness": 0,
     "text": "Path A\nfile/path_a.py"},
    {"type": "rectangle", "id": "l2b", "x": 230, "y": 570, "width": 290, "height": 65,
     "backgroundColor": "#a5d8ff", "strokeColor": "#1971c2", "roughness": 0,
     "text": "Path B\nfile/path_b.py"},

    // Converge point
    {"type": "rectangle", "id": "l4", "x": 60, "y": 720, "width": 300, "height": 65,
     "backgroundColor": "#eebefa", "strokeColor": "#9c36b5", "roughness": 0,
     "text": "Convergence Point\nfile/path_merge.py"},

    // Decision
    {"type": "diamond", "id": "dec", "x": 110, "y": 870, "width": 200, "height": 120,
     "backgroundColor": "#fff3bf", "strokeColor": "#fab005", "roughness": 0,
     "text": "condition?"},

    // Outcome branches
    {"type": "rectangle", "id": "yes", "x": -100, "y": 1080, "width": 260, "height": 65,
     "backgroundColor": "#ffc9c9", "strokeColor": "#e03131", "roughness": 0,
     "text": "Expensive Operation"},
    {"type": "rectangle", "id": "no", "x": 250, "y": 1080, "width": 220, "height": 65,
     "backgroundColor": "#b2f2bb", "strokeColor": "#2f9e44", "roughness": 0,
     "text": "Skip / Fast Path"},

    // --- ARROWS (bound by ID, auto-routed) ---
    {"type": "arrow", "x": 150, "y": 485, "startElementId": "l1", "endElementId": "l2a",
     "text": "Path A label", "strokeColor": "#1971c2"},
    {"type": "arrow", "x": 280, "y": 485, "startElementId": "l1", "endElementId": "l2b",
     "text": "Path B label", "strokeColor": "#1971c2"},
    {"type": "arrow", "x": 45, "y": 635, "startElementId": "l2a", "endElementId": "l4",
     "text": "data form", "strokeColor": "#9c36b5"},
    {"type": "arrow", "x": 375, "y": 635, "startElementId": "l2b", "endElementId": "l4",
     "text": "data form", "strokeColor": "#1971c2", "strokeStyle": "dashed"},
    {"type": "arrow", "x": 210, "y": 785, "startElementId": "l4", "endElementId": "dec",
     "strokeColor": "#2f9e44"},
    {"type": "arrow", "x": 150, "y": 990, "startElementId": "dec", "endElementId": "yes",
     "text": "True", "strokeColor": "#e03131"},
    {"type": "arrow", "x": 270, "y": 990, "startElementId": "dec", "endElementId": "no",
     "text": "False", "strokeColor": "#2f9e44"},

    // --- LAYER LABELS (left column, gray) ---
    {"type": "text", "x": -100, "y": 420, "text": "Layer 1\nEntry", "fontSize": 14, "strokeColor": "#868e96"},
    {"type": "text", "x": -210, "y": 570, "text": "Layer 2\nBackend", "fontSize": 14, "strokeColor": "#868e96"},

    // --- DATA FORM ANNOTATIONS (right column, orange) ---
    {"type": "text", "x": 570, "y": 440, "text": "Data form: Python bool", "fontSize": 14, "strokeColor": "#e8590c"},
    {"type": "text", "x": 570, "y": 590, "text": "Data form: JSON / arg", "fontSize": 14, "strokeColor": "#e8590c"},
    {"type": "text", "x": 570, "y": 740, "text": "Data form: Dataclass", "fontSize": 14, "strokeColor": "#e8590c"}
  ]
}

Color Palette (Quick Reference)

Rule: Same-role shapes get same colors. Limit to 3-4 fill colors per diagram.

Sizing Rules

Err on the side of too much space. Tight spacing is the #1 mistake — arrows and their labels get hidden when boxes are too close. When in doubt, double the gap you think you need. Diagrams that feel "too spread out" in your head almost always look right on screen.

CRITICAL: Arrow labels need ~120px of clear space between boxes to be visible. If an arrow has a text label (e.g. "auto deploy", "All pass"), the gap between the two connected boxes MUST be at least 150px. Arrows without labels still need 100px minimum.

Zone sizing rule: Calculate zone dimensions as: leftmost child x - 50 to rightmost child x + child width + 60 (horizontal), topmost child y - 55 to bottommost child y + child height + 60 (vertical). Always verify the zone fully wraps ALL children with visible padding on every side.

Arrow visibility test: Before finalizing, mentally check every labeled arrow — if the label text is longer than half the gap between boxes, increase the gap. Common offenders: "auto deploy", "rollback on failure", "All pass" — these labels are 80-150px wide and get clipped when gaps are <150px.

Layout Patterns

Vertical Flow (default for most diagrams)

Title (y = -40)

[Zone 1: y=0, height=260]
  [Box A: x=40]    [Box B: x=440]    [Box C: x=840]

[Zone 2: y=350, height=260]
  [Box D: x=40]    [Box E: x=440]

[Zone 3: y=700, height=260]
  [Box F: x=240]

Arrows flow top to bottom. Cross-layer arrows use dashed style.

Horizontal Pipeline

[Source] ──► [Transform 1] ──► [Transform 2] ──► [Output]
  x=40        x=440             x=840             x=1240

All at same y. Arrows flow left to right. Use 400px column spacing for labeled arrows, 340px for unlabeled.

Hub and Spoke

         [Consumer A]
              ▲
              │
[Producer] ──► [Event Bus] ──► [Consumer B]
              │
              ▼
         [Consumer C]

Central shape at (300, 300). Spokes at ~200px radius.

Data Flow Diagram (parameter threading, call chains)

Best for: tracing a parameter/request through architectural layers, showing data transformations at each boundary.

 Layer Labels        Main Flow Column          Side Annotations
 (left, gray)        (center, colored)         (right, orange)

  Layer 1      ┌─────────────────────┐         Data form: Python bool
  User API     │  Entry Point        │
               │  file/path.py       │
               └──────┬──────┬───────┘
                      │      │
              ┌───────┘      └────────┐
              ▼                       ▼
  Layer 2  ┌──────────┐    ┌──────────┐        Data form: JSON / bool
  Backend  │ Path A   │    │ Path B   │
           └────┬─────┘    └────┬─────┘
                │               │  (dashed = direct)
                ▼               │
  Layer 3  ┌──────────┐        │               Data form: Dataclass
  HTTP     │ Server   │        │
           └────┬─────┘        │
                └──────┬───────┘  (converge)
                       ▼
  Layer 4  ┌─────────────────────┐             Data form: IPC message
  Manager  │  Manager            │
           └──────┬──────────────┘
                  ▼
  Layer 5  ┌─────────────────────┐             Data form: Conditional
  Executor │  Executor           │
           └──────┬──────────────┘
                  ▼
               ◇ Decision? ◇
              / \
         True/   \False
            ▼     ▼
         [Yes]  [No]

Golden coordinates (validated):

Three-column structure:

• Left column (x < 0): Layer numbers + names in gray (#868e96, fontSize 14)
• Center column (x: 60–360): Flow boxes with ComponentName\nfile/path.py
• Right column (x: 570): Data form annotations in orange (#e8590c, fontSize 14)

Color by layer role:

• Blue (#a5d8ff/#1971c2): User-facing API layers
• Purple (#eebefa/#9c36b5): Internal processing layers
• Green (#b2f2bb/#2f9e44): Execution layer + "skip/success" outcomes
• Yellow (#fff3bf/#fab005): Decision nodes
• Red (#ffc9c9/#e03131): Expensive/dangerous operations
• Gray (#e9ecef/#868e96): Annotations, zones

Split and converge pattern:

• Split: Two arrows from one box, angled down-left and down-right
• Direct skip: Use strokeStyle: "dashed" for paths that skip layers
• Converge: Two arrows from different paths into one box below

"WHY" annotation box (first-principles context): Place a gray-background rectangle (top-right, x: 460) with 3-4 text items explaining the motivation. Use red stroke for the problem, green stroke for the solution.

Common Mistakes and Fixes

Quick Start Templates

"Draw me a diagram of X"

# 1. Read the code to understand components and connections
# 2. Read the design guide
mcp__excalidraw__read_diagram_guide()
# 3. Clear canvas and verify empty
mcp__excalidraw__clear_canvas()
# 4. Create everything in one batch
mcp__excalidraw__batch_create_elements(elements=[...])
# 5. Self-critique loop runs automatically (snapshot, validate, fix)
# 6. Present to user, then export if requested

"Quick diagram from description"

# For simple diagrams, Mermaid is fastest:
mcp__excalidraw__create_from_mermaid(
  mermaidDiagram="graph TD; A[Frontend] -->|REST| B[API]; B -->|SQL| C[Database]"
)

Export Options

After Diagram Delivery

Show this message once per conversation — only after the first successful diagram the user confirms they're happy with. Do not repeat on subsequent diagrams in the same session.

If you found this useful, I'd love your feedback! Feature requests, bug reports, or just a star: https://github.com/edwingao28/excalidraw-toolkit/issues/13