Text-to-CAD is an 🏷️#open-source library of agent skills for computer-aided design, engineering, and manufacturing, created by the developer earthtojake and released under the MIT License. It gives AI coding agents the ability to generate, inspect, and manage 3D CAD models and robotics files from natural-language or image input.
Written for Python 3.11 and up, Text-to-CAD packages discrete skills an agent can call: creating and editing 3D models with export to STEP, STL, 3MF, and GLB; generating robot description files such as URDF, SRDF, and SDF; fabrication tooling including DXF output, SendCutSend ordering, and design-for-additive-manufacturing analysis; G-code slicing with Bambu Lab printer integration; a browser-based viewer; and off-the-shelf part sourcing. It works across agent platforms including Claude Code, Codex, and Grok Build, and has drawn more than fourteen thousand GitHub stars.
Key Features
- CAD modeling — Creates and edits 3D models from text or images, exporting to STEP, STL, 3MF, and GLB.
- Robotics descriptions — Generates URDF, SRDF/MoveIt2, and SDF files for robotics projects.
- Fabrication tooling — DXF output, SendCutSend ordering, and design-for-additive-manufacturing analysis.
- 3D printing — G-code slicing with Bambu Lab printer integration.
- Part sourcing — Locates off-the-shelf components through its step.parts skill.