Text-to-CAD

Tracked

A collection of agent skills for CAD, robotics, and hardware design that let AI agents generate production-ready mechanical parts, 3D models, and robot descriptions from natural language.

Author earthtojake Open Sourced 2026-04-22 Last Commit Unknown

Overview

Text-to-CAD is an open-source collection of agent skills that give AI coding agents the ability to design physical hardware. Built by Jake (earthtojake), the skills bridge natural-language prompts with parametric CAD kernels and robotics model formats, so an agent can produce manufacturable geometry and validated robot descriptions end to end.

Key Features

  • Generates CAD geometry in formats such as STEP, STL, 3MF, DXF, and GLB using the build123d and OpenCascade kernels
  • Produces robotics description files including URDF, SDF, SRDF for mechanical assemblies and robot models
  • Structured as composable agent skills that plug into existing AI coding workflows
  • Targets mechanical engineering, robotics, and hardware design use cases from a single prompt

Use Cases

  • Turning a text spec into a parametric, manufacturable mechanical part
  • Generating robot and assembly models with valid URDF/SDF descriptions
  • Prototyping hardware designs inside an AI-assisted development loop

Technical Details

  • Built on the build123d parametric CAD library and OpenCascade geometry kernel
  • Outputs span manufacturing formats (STEP/STL/3MF/DXF) and web formats (GLB) alongside robotics formats (URDF/SDF/SRDF)
  • Distributed as agent skills, matching the SKILL.md capability module pattern used by modern AI coding agents