An open-source project called Text-to-CAD is drawing attention for wiring AI coding agents such as Claude Code and Codex directly into physical design, letting users generate real 3D CAD models, robot description files and 3D-printer instructions from plain-English prompts.
Open Source · text-to-cad
From Plain English to a Printed Part
An open-source harness wires AI coding agents like Claude Code and Codex straight into physical design — turning a prompt into real 3D CAD models, robot description files, and printer-ready G-code, all running locally.
8+
concrete prompt examples shipped with verification files
7-DOF
robot arm kinematics + GUI generated from a prompt
$0
free, local & offline, community-driven
End-to-end pipeline, one prompt
Prompt
plain English / image
→
CAD Model
STEP · STL · 3MF · GLB
→
G-code
via real slicer CLI
→
Printed Part
Bambu Lab print
A directory of agent "skills"
Extends existing coding agents rather than shipping a new model — installed with one command.
Core CAD → STEP
Browser CAD Viewer
DXF export
URDF robots
SRDF (MoveIt)
SDF simulators
G-code + Bambu print
SendCutSend validation
Implicit CAD (GLSL SDF)
Where it shines
Print-ready parts on the first attempt for some users
More practical than built-in CAD AI features
Simple designs need only a minor tweak or two
Collapses a brittle multi-tool pipeline
The caveats
Strong on simple shapes; needs detailed descriptions
Manual verification, often alongside OpenSCAD
Ambiguity (e.g. exact hole placement) trips it up
Larger, complex projects expected to hit edge cases
Released in stages since May 2026
An early but tangible step toward agentic AI reaching beyond software into physical, manufacturable design — advancing from v0.1.0 (Bambu + G-code) through the 0.3.x series with a faster viewer and plugin support.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…