Fusion-360-MCP-Server

AuraFriday/Fusion-360-MCP-Server

3.4

If you are the rightful owner of Fusion-360-MCP-Server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

Control Fusion 360 with any AI through Model Context Protocol (MCP).

Autodesk Fusion MCP Server Add-in

Control Fusion with AI !

This add-in for Autodesk Fusion connects to the Aura Friday MCP-Link server, making Fusion available as a remote tool that AI agents can control. Official Store Link

Watch Demo

Click above to watch AI create "Fusion Rocks!" in 3D

🚀 What's New

Latest: Thread-Safe Architecture (November 2025)

  • ✅ Rock-solid stability - All Fusion API calls now execute on main thread
  • ✅ Zero crashes - Work queue system prevents threading issues
  • ✅ Centralized logging - Thread-safe diagnostics from all components
  • ✅ Production ready - Handles heavy loads without instability

Python Execution + MCP Tool Integration

AI can now run arbitrary Python code directly inside Fusion with full access to:

  • ✅ Entire Fusion API (adsk.core, adsk.fusion, adsk.cam)
  • ✅ All loaded add-ins (access your custom add-ins automatically!)
  • ✅ SQLite databases (store and query design data)
  • ✅ Browser automation (open documentation, extract data)
  • ✅ User popups (show results, get input)
  • ✅ 500+ AI models via OpenRouter
  • ✅ Local AI models (offline inference)
  • ✅ Desktop automation (Windows control)

Example: AI analyzes your AirfoilTools add-in, finds the best airfoil from 1,538 profiles, stores results in SQLite, and shows a popup—all in one command!

What This Is

A Fusion add-in that gives AI unlimited access to Fusion through three powerful capabilities:

1. 🎯 Generic API Calls

Execute any Fusion API command without custom code:

fusion360.execute({
  "api_path": "design.rootComponent.sketches.add",
  "args": ["design.rootComponent.xYConstructionPlane"]
})

2. 🐍 Python Execution (NEW!)

Run arbitrary Python with full Fusion access:

fusion360.execute({
  "operation": "execute_python",
  "code": """
import adsk.core, adsk.fusion

# Create sketch
sketch = design.rootComponent.sketches.add(design.rootComponent.xYConstructionPlane)

# Store in database
mcp.call('sqlite', {
    'input': {'sql': 'INSERT INTO designs (name) VALUES (?)', 
              'params': [sketch.name]}
})

# Show popup
mcp.call('user', {
    'input': {'operation': 'show_popup', 
              'html': f'<h1>Created {sketch.name}!</h1>'}
})
"""
})

3. 🔗 MCP Tool Integration (NEW!)

Access 10+ built-in tools from Fusion:

  • SQLite databases
  • Browser automation
  • User popups
  • Python execution
  • AI models (local & cloud)
  • Desktop automation
  • And more!

🎯 What AI Can Do

Basic Operations

  • ✅ Create parametric sketches with precise dimensions
  • ✅ Add text and extrude it to 3D
  • ✅ Generate mounting holes, brackets, mechanical parts
  • ✅ Execute any Fusion API command

Advanced Workflows (NEW!)

  • 🔥 Access your custom add-ins - AI discovers and uses ANY loaded add-in
  • 🔥 Store design data - Save analysis results to SQLite
  • 🔥 Query databases - Find optimal designs from your data
  • 🔥 Show results - Display popups with charts, tables, forms
  • 🔥 Automate workflows - Combine Fusion + database + browser + AI
  • 🔥 Call AI models - Get design suggestions from 500+ models

Real-World Example

# AI analyzes AirfoilTools add-in (15,000+ users!)
# Finds best airfoil from 1,538 profiles
# Stores in SQLite
# Shows results in popup
# All in one Python command!

🛠️ Built-in MCP Tools

Everything you need. Nothing you don't.

These tools ship with MCP-Link and work immediately. No configuration, no API keys, no setup.

ToolDescription
🌐 BrowserAutomate Chrome: read, click, type, navigate, extract data
🧠 SQLiteDatabase with semantic search and embeddings
🐍 PythonExecute code locally with full MCP tool access
🤖 OpenRouterAccess 500+ AI models (free and paid)
🤗 HuggingFaceRun AI models offline (no internet needed)
📚 Context7Pull live documentation for any library
🖥️ DesktopControl Windows apps (click, type, read)
💬 UserShow HTML popups for forms, confirmations
🔗 RemoteLet external systems offer tools (like Fusion!)
🔌 ConnectorAdd any 3rd party MCP tools

Want more? Add any third-party MCP tools or build your own!

📦 Installation

Prerequisites

  1. Download MCP-Link Server
    Get the latest release: https://github.com/AuraFriday/mcp-link-server/releases/tag/latest

  2. Install the Add-In Official link: https://apps.autodesk.com/FUSION/en/Detail/Index?id=7269770001970905100

Optional manual install

2b. Clone This Repository

git clone https://github.com/AuraFriday/Fusion-360-MCP-Server.git

2c. Load as Fusion Add-in

  • Open Fusion
  • Press Shift+S to open Scripts and Add-Ins
  • Click the Add-Ins tab
  • Click the green + button next to "My Add-Ins"
  • Navigate to the cloned repository folder
  • Select the folder and click OK
  • Click Run to start the add-in

Configuration

The add-in auto-connects to the MCP server on startup. Check the TEXT COMMANDS window in Fusion to see connection logs.

🎬 Quick Start Examples

Example 1: Simple Sketch (Generic API)

# Create a sketch
fusion360.execute({
  "api_path": "design.rootComponent.sketches.add",
  "args": ["design.rootComponent.xYConstructionPlane"],
  "store_as": "my_sketch"
})

# Add a rectangle
fusion360.execute({
  "api_path": "$my_sketch.sketchCurves.sketchLines.addTwoPointRectangle",
  "args": [
    {"type": "Point3D", "x": 0, "y": 0, "z": 0},
    {"type": "Point3D", "x": 10, "y": 5, "z": 0}
  ]
})

Example 2: Python with Database (NEW!)

fusion360.execute({
  "operation": "execute_python",
  "code": """
import adsk.core, adsk.fusion

# Create mounting plate
sketch = design.rootComponent.sketches.add(design.rootComponent.xYConstructionPlane)
lines = sketch.sketchCurves.sketchLines
lines.addTwoPointRectangle(
    adsk.core.Point3D.create(0, 0, 0),
    adsk.core.Point3D.create(10, 8, 0)
)

# Add mounting holes
circles = sketch.sketchCurves.sketchCircles
for x, y in [(1, 1), (9, 1), (1, 7), (9, 7)]:
    circles.addByCenterRadius(adsk.core.Point3D.create(x, y, 0), 0.25)

# Store in database
mcp.call('sqlite', {
    'input': {
        'sql': 'INSERT INTO parts (name, holes) VALUES (?, ?)',
        'params': [sketch.name, 4],
        'database': 'designs.db',
        'tool_unlock_token': '29e63eb5'
    }
})

print(f'Created {sketch.name} with 4 mounting holes')
"""
})

Example 3: Access Custom Add-ins (NEW!)

fusion360.execute({
  "operation": "execute_python",
  "code": """
import sys

# Find loaded add-ins
addins = [name for name in sys.modules.keys() if 'addin' in name.lower()]
print(f'Found {len(addins)} add-ins')

# Access AirfoilTools (if loaded)
if 'AirfoilTools' in str(addins):
    airfoil_main = sys.modules['...AirfoilTools_py']
    foildb = airfoil_main.foildb2020.Foildb2020()
    
    # Find best airfoil
    best = max(foildb, key=lambda x: x['clcd'])
    print(f'Best L/D ratio: {best["clcd"]:.2f}')
    
    # Store in database
    mcp.call('sqlite', {
        'input': {
            'sql': 'INSERT INTO airfoils (name, ld_ratio) VALUES (?, ?)',
            'params': [best['Foil_fn'], best['clcd']],
            'database': 'airfoils.db',
            'tool_unlock_token': '29e63eb5'
        }
    })
"""
})

🏗️ Architecture

How It Works

  1. Add-in loads and connects to local MCP-Link server
  2. Registers itself as a fusion360 tool via reverse connection
  3. AI sends commands with three operation types:
    • Generic API calls (simple operations)
    • Python execution (complex workflows)
    • MCP tool calls (cross-tool integration)
  4. Generic handler resolves paths, constructs objects, executes API calls
  5. Python executor runs code with TRUE INLINE access to Fusion environment
  6. Results returned to AI with full context and type information

Components

  • MCP-Link Server - Routes AI requests to the add-in
  • Generic API Handler - Dynamically executes any Fusion API call
  • Python Executor (NEW!) - Runs arbitrary Python with full access
  • MCP Bridge (NEW!) - Allows Python to call other MCP tools
  • Context Manager - Stores intermediate objects for multi-step operations
  • Enhanced Reporting - Provides detailed success/error information
  • Thread-Safe Proxy (NEW!) - Ensures all Fusion API calls happen on main thread
  • Centralized Logger (NEW!) - Thread-safe logging from any component

🎨 Supported Patterns

API Paths

"design.rootComponent.sketches.add"              # Access via app object
"$sketch.sketchCurves.sketchCircles.addByCenterRadius"  # Use stored objects
"adsk.core.Point3D.create"                       # Full module paths
"adsk.core.ValueInput.createByString"            # Static methods

Object Construction

{"type": "Point3D", "x": 5, "y": 10, "z": 0}    # Construct from dict
"$my_sketch.originPoint"                         # Reference stored object
0                                                 # Enum as integer

Context Storage

"store_as": "my_object"                          # Store result as $my_object
"$my_object.someMethod"                          # Use in next command

Python Execution (NEW!)

"operation": "execute_python"                    # Run Python code
"code": "import adsk.core\n..."                  # Python code string
"session_id": "my_session"                       # Persistent session
"persistent": true                               # Variables persist

MCP Tool Calling (NEW!)

"operation": "call_tool"                         # Call another MCP tool
"tool_name": "sqlite"                            # Tool to call
"arguments": {"input": {...}}                    # Tool arguments

💡 Use Cases

AI Design Assistants

  • Natural language to CAD: "Create a mounting bracket with 4 holes"
  • Design exploration: "Try 10 variations of this part"
  • Parameter optimization: "Find the best dimensions for strength"

Data-Driven Design (NEW!)

  • Query databases: "Find the best airfoil for Re=500,000"
  • Store analysis: "Save this design's mass and volume"
  • Compare designs: "Show me all brackets lighter than 100g"

Workflow Automation (NEW!)

  • Multi-tool workflows: Create design → Store in DB → Show popup
  • Access custom add-ins: Use your existing add-ins via AI
  • Batch operations: Process 100 designs automatically

Educational Tools

  • Learn Fusion API through AI guidance
  • Interactive tutorials with instant feedback
  • Explore aerodynamics with AirfoilTools integration

CAM Automation

  • AI-driven toolpath generation
  • Optimize cutting parameters
  • Batch process manufacturing jobs

🎯 Real-World Example: AirfoilTools Integration

Scenario: You have the AirfoilTools add-in (15,000+ users) loaded in Fusion.

AI can:

  1. Discover the add-in automatically
  2. Access 1,538 airfoil profiles
  3. Find the best airfoil (L/D ratio: 505.02!)
  4. Store results in SQLite
  5. Create geometry in Fusion
  6. Show results in popup

All without any custom integration code!

See for the complete demo.

📚 Documentation

  • - 10 comprehensive tests
  • - Full capabilities
  • - Working examples
  • - Real-world integration

🔧 Technical Details

Core Architecture

  • Built with Fusion's native Python API
  • Uses Server-Sent Events (SSE) for MCP connection
  • JSON-RPC protocol for command execution
  • TRUE INLINE Python execution (no sandboxing)
  • Zero overhead on Fusion performance
  • All operations run locally (no cloud dependencies)
  • Auto-reconnection with exponential backoff
  • Native messaging protocol for server discovery

Thread Safety & Stability

  • Main Thread Execution: All Fusion API calls execute on Fusion's main UI thread
  • Work Queue System: Background threads queue work items for safe main thread processing
  • Custom Event Handler: Uses Fusion's CustomEvent system to safely trigger API calls
  • Centralized Logging: Thread-safe log buffer ensures reliable diagnostics from any thread
  • Crash Prevention: Robust architecture prevents crashes even under heavy load

⚠️ Limitations

  • Enum values must be passed as integers (e.g., 0 for NewBodyFeatureOperation)
  • Some complex objects need specific construction patterns
  • Context is lost when add-in is reloaded (working as intended for session isolation)
  • Python execution has FULL system access (use responsibly!)

🚀 Future Enhancements

  • Semantic search integration for auto-documentation
  • Batch operation optimization
  • Undo/redo helpers
  • Extended enum handling
  • More intuitive object construction patterns
  • Script library for common operations
  • Visual workflow builder

🤝 Contributing

Want to contribute? PRs welcome! This is the future of AI-powered CAD.

Ideas for Contributors

  • Add example workflows
  • Improve error messages
  • Create tutorial videos
  • Test with different add-ins
  • Build integration with other CAD tools

📄 License

Proprietary - See LICENSE file for details

👤 Author

Created by AuraFriday
With 15,000+ users trusting previous Fusion add-ins

🔗 Links

❓ FAQ

Q: Does this work with my existing add-ins?
A: Yes! AI can discover and use ANY loaded add-in automatically.

Q: Do I need to write custom integration code?
A: No! The generic API handler works with everything.

Q: Can AI access my databases?
A: Yes! Use the SQLite tool to store and query design data.

Q: Is this secure?
A: Python execution has full system access. Only run code you trust.

Q: Does this work offline?
A: Yes! All operations run locally. Optional: use local AI models.

Q: Can I use this with Claude/ChatGPT/etc?
A: Yes! Works with any AI that supports MCP protocol.

Q: Is it stable? Will it crash Fusion?
A: Very stable! Thread-safe architecture ensures all API calls happen on main thread, preventing crashes.

Q: Can I use this in production?
A: Yes! The add-in is production-ready with robust error handling and stability features.

🌟 Star This Project!

If you find this useful, please star the repository and share with the Fusion community!

Questions? Open an issue or check the documentation in

This is the future of AI-powered CAD. 🚀