Rhinocommon

You run a dedicated MCP server that provides RhinoCommon API documentation to assist Rhino plugin developers. It powers Claude Code to reference the exact RhinoCommon API during code generation, speeding up plugin development and reducing API-usage errors.

How to use

To use this MCP server with Claude Code, you register a local MCP endpoint that Claude Code can talk to. Once registered, you can ask Claude Code to search RhinoCommon concepts, fetch class details, and retrieve code examples. The server exposes a set of tools that Claude Code invokes behind the scenes to locate and present precise API information, which Claude Code then uses to generate correct RhinoCommon code for your plugin.

Typical usage patterns include searching for a class like NurbsSurface, inspecting the class API to see constructors and members, and requesting concrete code examples that show how to instantiate and use that API in a Rhino plugin. You can also ask for specific methods or operations, and Claude Code will guide you to the correct API usage with verified details.

How to install

Prerequisites you need before installing this MCP server: Docker (for containerized runs) and Python (for local runs). You will also need Git to clone the repository and Pip to install Python dependencies.

Step 1 — Clone the project and navigate into it.

git clone https://github.com/voidbox-ai/rhinocommon-mcp.git
cd rhinocommon-mcp

Install via Docker (recommended for quick start)

Step 2 — Start the crawling and MCP server services using Docker. This runs both the document crawler and the MCP server in containers.

git clone https://github.com/voidbox-ai/rhinocommon-mcp.git
cd rhinocommon-mcp

# Run document crawling and MCP server in background containers
docker-compose up -d

Install without Docker (local Python path)

Step 2 — Install dependencies and prepare the server locally for a direct Python run.

cd scraper
pip install -r requirements.txt
python scraper.py --version 8

cd ../server
pip install -r requirements.txt

Register MCP endpoints for Claude Code

Step 3 — Connect Claude Code to the MCP server. You can configure Claude Code to run the MCP server inside Docker or as a local Python process.

# VSCode Claude Code configuration
# command: docker
# args: ["exec", "-i", "rhinocommon-mcp-server", "python", "/app/server/mcp_server.py"]

# Claude Desktop configuration (example with absolute path)
# command: python
# args: ["/절대/경로/rhinocommon-mcp/server/mcp_server.py"]

Test the connection

After configuration, restart Claude Code and test a simple query to ensure the MCP server responds correctly. For example, ask Claude Code to provide information about the RhinoCommon Point3d class. If the MCP tools respond with class details and code examples, your connection is working.

Notes and troubleshooting

If you encounter issues, verify that the MCP server is reachable from Claude Code and that the crawler documentation has completed loading (especially when using Docker). Check container statuses with docker ps and ensure the rhinocommon-mcp-server container is running. If you switch between Docker and local Python runs, update the client configuration accordingly and restart Claude Code.

Usage pattern overview

Tools available:
- search_rhinocommon: search for classes or members by name
- get_class_details: fetch all methods, properties, and constructors for a class
- get_code_examples: retrieve practical code samples

Available tools

search_rhinocommon

Search for RhinoCommon classes, methods, and members by name to locate the API surface you need.

get_class_details

Fetch all methods, properties, and constructors for a specific RhinoCommon class to understand its full API surface.

get_code_examples

Provide practical code samples showing how to use a RhinoCommon API in typical plugin scenarios.