CadQuery: Building 3D Models With Python Code

The open source project CadQuery was pushed to the front page: write 3D CAD models with Python scripts—parametric, version-controllable, and surprisingly smooth to pair with LLMs.

The Benefits of Code-Based Modeling

CadQuery's idea is to turn 3D modeling into writing code: dimensions are variables, structures are functions, changing one parameter recomputes the whole model in cascade, and the model file goes straight into git for version control. For engineers, this is far more comfortable than dragging in a graphical interface—reuse, batch generation, and automated drawing output are all effortless. It's been recommended repeatedly on HN for years, and this resurgence has a new reason: code-form models happen to be a format LLMs can read and write, so the "modeling by voice" workflow suddenly becomes realistic.

The Room for Imagination With AI

The liveliest direction in the comments is exactly this: have the model generate CadQuery scripts—describe an "enclosure with four mounting holes," and the code comes out, parameters adjustable, errors diffable—far more reliable than generating mesh files directly, because the code carries design intent, not just shape. Of course, the boundaries are clear too: LLMs' understanding of spatial relationships is still unreliable, and generating complex assemblies is basically a disaster; for now the sweet spot is simple parts and templated variants. But the path of "text to parametric solid" looks more like the future of CAD than "text to triangle mesh."

via: Hacker News