Setting up Clients

Overview

Morfium supports multiple AI service providers as long as models are trained to execute tools.

Supported Platforms

Configure Morfium with your preferred AI client. Each platform has specific setup requirements.

LM Studio

Local AI models with MCP support
  1. Install LM Studio and ensure you have a model loaded
  2. Create or edit your MCP configuration file
  3. Add the Blender server configuration:
    {
  "mcpServers": {
    "blender": {
      "transport": "http", 
      "url": "http://localhost:12097/mcp"
    }
  }
}
  4. Configure LM Studio to use MCP by enabling the MCP integration in settings
  5. The AI model can now interact with Blender through the Morfium MCP server

Claude Code

Official Claude CLI with full MCP integration

Local Setup (Same System):

claude mcp add --transport http blender http://localhost:12097/mcp

WSL Setup (Blender on Windows, Claude Code in WSL):

# Step 1: Enable "Allow Network Access" in Blender addon preferences

# Step 2: Find your Windows host IP
ip route | grep default | awk '{print $3}'

# Step 3: Use that IP (e.g., 172.26.208.1)
claude mcp add --transport http blender http://172.26.208.1:12097/mcp

# Step 4: Verify connection
claude --debug
# Look for: "MCP server 'blender': Connection attempt completed ... status: connected"

VS Code + GitHub Copilot

GitHub Copilot with MCP extension

Workspace Configuration:

  1. Create .vscode/mcp.json in your workspace (or run MCP: Add Server command)
  2. Add the Blender server configuration:
    {
  "servers": {
    "Blender": {
      "url": "http://localhost:12097/mcp"
    }
  }
}
  3. Connect to the server:
    • Open GitHub Copilot Chat
    • Click the "Configure Chat" button (settings icon)
    • Select "MCP Servers"
    • Find "Blender" and click "Start" to connect

Global Configuration:

  • Run MCP: Open User Configuration command
  • Add the same server configuration to your user mcp.json
  • Follow the same steps to start the server in Copilot Chat

Cursor

AI-powered code editor with built-in MCP
  1. Open Cursor Settings (Cmd/Ctrl + ,)
  2. Search for "MCP"
  3. Add server configuration:
    {
  "mcpServers": {
    "blender": {
      "transport": "http", 
      "url": "http://localhost:12097/mcp"
    }
  }
}

Claude Desktop

Native Claude app with stdio-based MCP

Claude Desktop requires stdio-based MCP servers. Use the official npm package to connect.

  1. Install Node.js and npm:
    • Download and install Node.js from nodejs.org
    • This will also install npm (Node Package Manager)
    • Verify installation: node --version and npm --version
  2. Configure Claude Desktop:
    • Windows: Edit %AppData%\Claude\claude_desktop_config.json
    • macOS: Edit ~/Library/Application Support/Claude/claude_desktop_config.json
  3. Add Configuration:
    {
  "mcpServers": {
    "blender": {
      "command": "npx",
      "args": ["-y", "morfium-blender-mcp"]
    }
  }
}
  4. Connect to Blender:
    • Start Blender and click the Mcp menu
    • Click "Start Server"
    • Restart Claude Desktop
    • Look for the plug icon and verify "blender" is connected

That's it! No file paths or manual setup needed.

## Next Steps Once you have at least one client configured: 1. Explore the [Options](options.html) to fine-tune generation parameters 2. Start with simple prompts to understand your model's capabilities 3. Experiment with different providers to find your preferred workflow