text-to-cad-mcp
MCP server that generates CAD parts and assemblies from text descriptions for any MCP-compatible AI client.
What it does and how it connects
gNucleus Text To CAD MCP server connects MCP clients to the gNucleus API, so they can generate CAD parts or assemblies from text input using gNucleus's generative models. It exposes one MCP tool, text_to_cad, and works with any MCP-compatible client.
Setup needs Python 3.7 or later, a gNucleus developer account, and a gNucleus API key in a .env file, with an organization ID for enterprise users. Each account starts with 200 free API credits, per the README. Prompts work in any language, though the README says English generally works better.
Generated results include the design spec with key parameters and a description, plus a shared URL with a 3D display viewer that expires in 24 hours. Example prompts cover blocks, spur gear shafts, elbow flanges, tapered roller bearings, and differential gear boxes.
Alternatives
Compare side by side- text-to-cad (CAD Skills)
An agent skills framework that gives AI coding agents parametric CAD superpowers through the build123d Python library.
- CADAM
An open source text-to-CAD web application that generates 3D models from natural language, powered by OpenSCAD running in the browser.
- cad-agent
An AI-driven CAD modeling agent with a visual feedback loop built on build123d and MCP for 3D-printable parts.
- CQAsk
An open source LLM-powered CAD generation tool that produces 3D models from natural language using CadQuery.
- cadquery-mcp-server
An MCP server that exposes CadQuery-based CAD generation and automated model validation tools to AI assistants.
Sources
Not verified
- No GitHub releases exist, so latest release fields are null.
- Dollar pricing (no dollar price is stated on the repo; only the 200 free starting credits).
- Real user voices (no real linked post read for this page).
- A named individual builder beyond the gNucleus brand
Last verified November 6, 2026. Listings are informational. The evidence score measures public, verifiable evidence and is not an endorsement or a compatibility claim.