Rhino connects to an AI assistant through the Model Context Protocol (MCP) with the native
RhinoMCP
and
mcpstart
commands. Where Vibe Coding your Tools in Rhino turns a single conversation into a script and a command alias, this tutorial goes further: we’ll use an AI coding assistant and a developer tool such as Visual Studio Code to build a real Rhino plug-in, from first prototype to something you can package and share.
What You’ll Need
- Rhino 9 for Windows or Mac.
- An MCP-capable AI chat client, such as Claude Desktop, that can connect to a local MCP server.
- Visual Studio Code (or another code editor) to hold the plug-in project.
- No prior plug-in development experience required. The assistant writes and edits the code for you; some familiarity with C# or Python is helpful, but not necessary, if you want to read along.
Connect Rhino to an AI Assistant
NOTE TO SELF: Reuse/link the connection steps from Vibe Coding your Tools in Rhino once that section is finalized, instead of duplicating it here.
Prototype the Plug-in Idea in Conversation
NOTE TO SELF: Walk through describing the plug-in’s intended behavior to the assistant and iterating on a working prototype in the live Rhino session, the same way everyday-tools iterates on the speaker grid script.
Scaffold the Plug-in Project
NOTE TO SELF: Cover generating a real Rhino plug-in project (e.g. via the yak / Visual Studio / VS Code plug-in templates) with the assistant’s help, and moving the prototyped logic into it.
Build and Test in Rhino
NOTE TO SELF: Cover loading the plug-in into Rhino for testing (PlugInManager, debug attach from the editor), and iterating with the assistant when something doesn’t work.
Package the Plug-in
NOTE TO SELF: Cover packaging with
PackageManager
/ Yak for distribution. See How to Use Scripts and Plugins for existing packaging guidance to draw from.
Where to Go from Here
- Prefer a lighter-weight approach first? See Vibe Coding your Tools in Rhino for turning a single script into a command and toolbar button, no packaging required.
- Remember to close the
mcpstart
bridge when you’re not actively using it — there’s no reason to leave an AI assistant with live code execution access to Rhino running in the background.
