FraktalMCP
AI for Grasshopper
Tell Grasshopper what you want in plain English. It builds the definition for you.
- Open-source, MIT licensed
- Works with Claude, Gemini, and local models
- Text to Grasshopper definitions
- Script generation (C# & Python)
- Definition analysis and debugging
# FraktalMCP
$ fraktalmcp connect --rhino
Connected to Rhino 8
$ fraktalmcp generate \
"parametric tower with twisted floors"
Generated 14 components, 23 connections
Definition saved to canvasThree Steps
Describe
Type what you want in natural language: "Create a parametric facade with hexagonal cells that respond to solar exposure."
Generate
AI selects the right Grasshopper components, wires them together, sets initial parameters, and builds a working definition.
Refine
Tweak sliders, ask for changes in plain text, or let AI analyze your definition and suggest improvements. Iterate until perfect.
Who It's For
Skip the wiring
Describe your design intent and get a working Grasshopper definition. No more hours of dragging and connecting components.
- Natural language to definition
- Complex pattern generation
- Rapid prototyping
Accelerate your workflow
Let AI handle the repetitive wiring while you focus on design logic and optimization strategies.
- Script generation (C# & Python)
- Definition analysis & debugging
- Automated parameter setup
Learn by example
Describe what you want to build and see how an expert would wire it. Understand parametric design through generated examples.
- Step-by-step explanations
- Best practices built in
- Experiment rapidly
What It Can Do
From simple geometry to complex multi-plugin workflows.
Text to Geometry
AvailableDescribe shapes, patterns, or structures in plain English. Get working Grasshopper definitions with correct wiring.
Definition Analysis
AvailablePaste an existing definition. AI explains what it does, identifies problems, and suggests optimizations.
Script Generation
AvailableNeed a C# or Python script component? Describe the logic, get production-ready code with error handling.
Multi-Model Support
AvailableWorks with Claude Opus 4.6, Sonnet 4.5, Gemini 2.5, local models via LM Studio/Ollama, or any MCP-compatible endpoint.
Plugin Awareness
Coming SoonUnderstands Ladybug, Karamba, Kangaroo, and 40+ other Grasshopper plugins. Routes to the right components.
Visual Debugger
Coming SoonHighlights data type mismatches, missing connections, and performance bottlenecks in existing definitions.
What Can You Ask?
Create a parametric facade with hexagonal cells that vary by solar exposure
Attractor-based hex grid + Ladybug solar analysis
Optimize this structure for minimum material
Karamba analysis + Galapagos evolutionary solver
Make roof panels follow a double-curved surface
Kangaroo planarization on subdivided quad mesh
Supported AI Models
FraktalMCP connects to any AI model that speaks your language. Use cloud providers or run fully local.
Claude Opus 4.6
AnthropicBest for complex multi-step definitions and analysis
Claude 4.5 Sonnet
AnthropicExcellent at code generation with fast response
Gemini 2.5 Pro
Google DeepMindStrong reasoning for structural optimization
GPT-4.1
OpenAIReliable general-purpose generation
LM Studio
LocalUser-friendly local model management with GPU acceleration
Ollama
LocalRun open-source models privately on your machine
Any MCP-compatible
CustomConnect to any server that supports the MCP protocol
How It Connects
FraktalMCP sits between your AI model and Rhino/Grasshopper. It translates natural language into component operations.
Manual vs FraktalMCP
| Manual | FraktalMCP | |
|---|---|---|
| Wiring | Drag, connect, adjust | Describe in text |
| Scripts | Write from scratch | Generated for you |
| Errors | Find at runtime | Caught before run |
| Speed | Minutes per change | Seconds per change |
Open Source
Free, open-source, MIT licensed. Built by architects for architects. Contributions welcome.
$ pip install fraktalmcp Frequently Asked Questions
What is the Model Context Protocol?
MCP is an open standard by Anthropic that lets AI assistants communicate directly with external tools. FraktalMCP implements this protocol specifically for Grasshopper, allowing Claude or other AI to read, write, and execute parametric definitions.
Do I need coding experience?
No. FraktalMCP translates natural language into Grasshopper components. You describe what you want in plain English, and it generates the definition. Advanced users can still write custom scripts when needed.
Which Grasshopper plugins are supported?
FraktalMCP supports all native Grasshopper components plus popular plugins including Ladybug, Honeybee, Kangaroo, Lunchbox, and Weaverbird. We're adding more each month.
Is it safe to use with client projects?
Yes. FraktalMCP runs entirely locally on your machine. No project data is ever sent to external servers. The AI only receives the component structure, not your actual geometry data.
How do I install FraktalMCP?
Simply run 'pip install fraktalmcp' in your terminal. Then connect it to your preferred AI assistant (Claude, GPT, etc.) using the configuration guide in our documentation.
What Developers Say
“FraktalMCP saved us weeks of Grasshopper wiring on a complex parametric facade. The AI just gets what architects mean.”
“I've been scripting GH components for years. FraktalMCP generates better definitions than most junior programmers.”
“The open-source approach won us over. We deployed it across our 40-person office in a day.”
Prompt → Geometry
Type a description. Get a working Grasshopper definition.