Devotd Github
NewAI agents

laszapens/atta — A parametric CAD kernel for LLMs

laszapens/atta is a Python CAD tool designed so that language models can build specification-based geometry and hand off the resulting artifacts through measurement and three-state validation.

Repositorylaszapens/atta ↗

laszapens/atta is a Python CAD tool designed so that language models can build specification-based geometry and hand off the resulting artifacts through measurement and three-state validation.

GitHub description: Parametric CAD kernel for LLMs: spec-driven geometry, measured facts, three-state verification, Blender-backed delivery

stars
★ 23
Forks
0
Language
Python
License
MIT
Read original source ↗View on GitHub ↗

laszapens/atta is a parametric CAD kernel for language models. The latest public release version is not given in FACTS, so this introduction covers the features described in the repository README. Atta is designed to build parametric geometry from a reconstructable specification, and to measure dimensions and topology before rendering. Instead of guessing at a shape from a picture alone, you check computed information.

Measurements are kept as "measurement facts", and validation is recorded with three states — pass, degraded, and fail — together with a receipt. According to the README, results that do not pass are restricted from being handed off. Blender is not required for the core geometry computation and is used only optionally for rendering and some exports. A command-line tool and a tool for agents are provided as well. It is written in Python and uses the MIT license. It is a tool that ties the process of handing a language model's shape over as a real design artifact to measurement and validation.

The repository records 21 stars, and no latest release is given. The research record scores novelty 8.0, technical 8.0, quality 8.0 and momentum 3.5. No real-world usage cases or performance figures are presented in FACTS, so the effect in actual design work cannot be judged from this alone. What can be confirmed right now is the functionality and design direction described in the README, and the practical usefulness of the tool can only be estimated once additional evidence is added.