AI Briefing
KO

Amagine3D Releases Parametric CAD Generation Agent

·2026.09.02 18:30

Key point

Amagine3D has released an open-source agent that generates editable parametric CAD sources from natural language and dimensions.

1 / 3

Details

Amagine3D is an open-source project that takes natural language requirements, reference images, and key dimensions as input to generate parametric CAD designs for hardware cases, preserving them as editable Python(build123d) source code.

Parametric CAD Adopted Instead of Meshes

While most existing image-based 3D generation tools output mesh formats that are difficult to edit, Amagine3D adopts a parametric approach that manages dimensions as variables. This allows for shape recalculation by simply changing parameter values in the source code without calling an LLM when fine adjustments, such as wall thickness or screw hole positions, are needed.

3D-Native Agent Architecture

The project uses a '3D-native agent' architecture centered on design state.

  • Autonomous Inner Loop: Generates candidate designs and checks for assembly interference, motion paths, and output tolerances in a real geometric environment.
  • Commit Stage: Only designs that pass checks are saved as new baselines, and user approval may be required when modifying existing structures.

Deliverables and Tech Stack

Generated designs can be exported in STEP/STL format for single-color models and 3MF format for multi-color models. Geometric operations are based on the Open CASCADE Technology and CadQuery OCP kernels, with Three.js used for browser previews. It can be run locally with Node.js and Python 3.10~3.13, and agent functionality is enabled by configuring LLM API keys.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.