fusion
A model context protocol (MCP) server for Autodesk Fusion that provides resources and tools from ADSK to an AI client such as Claude or Cursor.
claude mcp add --transport stdio joe-spencer-fusion-mcp-server python -m fusion_mcp_server \ --env MCP_SDK_VERSION="latest" \ --env FUSION_ADDIN_PATH="Path to Fusion 360 add-in folder or leave as empty if auto-discovered" \ --env FUSION_PYTHON_ENV="Fusion 360 Python environment identifier"
How to use
This Fusion 360 MCP server exposes Fusion 360 functionality to MCP-enabled AI agents. The server runs as a Fusion 360 add-in and bridges CAD operations, design data, and prompts to external agents via the MCP protocol. You can query resources like the active document, design structure, and parameters; invoke tools such as creating sketches, adding parameters, or displaying message boxes; and access specialized prompts to guide CAD tasks. To test locally, use the included client.py utility to exercise the API and verify connectivity. The server’s MCP implementation supports both HTTP SSE communication and a file-based fallback, ensuring compatibility in environments where direct network access is restricted.
How to install
Prerequisites:
- Autodesk Fusion 360 installed
- Python 3.7+ available inside Fusion 360’s Python environment (or accessible for testing via the provided scripts)
- MCP Python SDK installed in Fusion 360’s Python environment (pip install "mcp[cli]")
Step 1: Install MCP in Fusion 360’s Python Environment
-
Using the installer script (recommended): python install_mcp_for_fusion.py This script discovers Fusion 360 Python installations and installs the MCP package with CLI extras in each one.
-
Manual installation (if the installer script cannot be used):
- Locate Fusion 360’s Python executable (typically under Autodesk\webdeploy\production[version]\Python)
- Run: "[Fusion Python Path]\python.exe" -m pip install "mcp[cli]"
Step 2: Install the Fusion 360 Add-in
- In Fusion 360, go to Tools -> Add-Ins -> Scripts and Add-Ins
- Click the green + button in the My Add-Ins tab
- Browse to and select the MCPserve folder from this repository
- Click Open to add it
- Select the MCPserve add-in and click Run to start it
Step 3: Run and Test
-
Use the included client.py to test the server: python client.py --test-connection
-
Test specific capabilities: python client.py --test-message-box python client.py --list-resources python client.py --list-tools python client.py --list-prompts
Additional notes
Notes and tips:
- Fusion 360 runs the MCP server as an add-in; external startup is not required beyond enabling the add-in within Fusion 360.
- The MCP Python SDK is used inside Fusion 360’s Python environment; ensure the environment used by Fusion 360 has the mcp[cli] extras installed.
- If you encounter connectivity issues, leverage the file-based MCP communication fallback by placing commands in the mcp_comm directory as described in the README.
- The server exposes resources (fusion://active-document-info, fusion://design-structure, fusion://parameters), tools (message_box, create_new_sketch, create_parameter), and prompts (create_sketch_prompt, parameter_setup_prompt). Use the client utility to explore and verify each capability.
- If running in a restricted environment, prefer the HTTP SSE endpoint at http://127.0.0.1:3000/sse, otherwise rely on the file-based method for robustness.
Related MCP Servers
mcp-vegalite
MCP server from isaacwasserman/mcp-vegalite-server
github-chat
A Model Context Protocol (MCP) for analyzing and querying GitHub repositories using the GitHub Chat API.
nautex
MCP server for guiding Coding Agents via end-to-end requirements to implementation plan pipeline
pagerduty
PagerDuty's official local MCP (Model Context Protocol) server which provides tools to interact with your PagerDuty account directly from your MCP-enabled client.
futu-stock
mcp server for futuniuniu stock
mcp -boilerplate
Boilerplate using one of the 'better' ways to build MCP Servers. Written using FastMCP