Rhino MCP Platform connects an external AI assistant to Rhino and Grasshopper through the Model Context Protocol. The setup brings together your Rhino installation and the connection configuration for your chosen assistant. Depending on your Rhino build, this uses the built-in connection tools or a separate plugin and connector.

This is separate from the AI panel inside Rhino. Our RhinoAI and RhinoMCP guide explains the distinction.

1. Choose Your Assistant and Installation Route

Start with Rhino installed and licensed, then choose the connection route documented for your Rhino build and AI assistant.

For Rhino 9 builds with MCPConnect, install your chosen AI application and run that command in Rhino. Find the application in the connection dialogue and choose Install to add the Rhino MCP server configuration, then follow the application's instructions to load the updated configuration. This configures the connection rather than installing the AI application. McNeel’s MCPConnect instructions also explain manual configuration for other clients.

For the separate package-based route, open McNeel’s getting-started guide. It provides separate routes for Claude Desktop, Claude Code, GitHub Copilot, OpenAI Codex, Gemini CLI and local models. The Claude Desktop guide currently labels its direct Rhino-plugin installation link for Rhino 8. Use the requirements and installation instructions supplied with the current plugin and client guide for your setup.

2. Package-Based Setup for Claude Desktop

For this installation route, use McNeel’s connector guide to download connector.mcpb. In Claude Desktop, open Settings → Extensions → Advanced settings, choose Install Extension and select the downloaded connector. Confirm installation when prompted.

The same guide provides a Rhino 8 installation link for Rhino-MCP-Platform and an alternative installation method through Claude using Yak, Rhino’s package manager. Use the published instructions for your setup, including restarting Rhino after the plugin is installed.

For other assistants, follow their individual setup pages rather than applying the Claude Desktop extension steps. Each route explains how that client connects to the Rhino MCP service.

3. Make a Small First Test

Open a fresh Rhino document and a new assistant chat, then follow McNeel’s Try It Out guide. Begin with a simple object so you can see which Rhino document receives the result and inspect the geometry directly.

Once that connection is established, try a small task in a copy of one of your models. Keep the request specific and review the resulting objects before continuing. This gives you a clear way to learn what the available tools do within your own modelling workflow.

4. Keep the Connection Easy to Check

If the result appears elsewhere than expected, check which Rhino document and window are active. Review object visibility and the view before repeating an operation, so you can distinguish an unseen result from an operation that has yet to complete.

Use McNeel’s troubleshooting guide for connection diagnostics. For a local-model route, begin with its local-model setup guide. Choose an AI service and permissions that fit your organisation’s approach to project information.

Develop Your Workflow

After the initial setup, our guide to working with VisualARQ, LandsDesign and plugin data via MCP explores a more specialised use. For repeatable instructions, see turning your office manual into reusable AI instructions.

Keep the developer’s setup guide as your installation reference as the platform evolves. This article provides the sequence and context; the linked client instructions supply the current installation details.