Introduces a multi-agent CAD generation pipeline that uses programmatic geometric validation from the OpenCASCADE kernel to iteratively fix dimensional errors.
March 30, 2026
Original Paper
CADSmith: Multi-Agent CAD Generation with Programmatic Geometric Validation
arXiv · 2603.26512
The Takeaway
Rather than relying on lossy visual feedback or pure token prediction, this method closes the loop with exact physical measurements (volume, solid validity). This shift from 'looking like' to 'validating as' geometry results in a 100% execution rate and near-perfect IoU scores.
From the abstract
Existing methods for text-to-CAD generation either operate in a single pass with no geometric verification or rely on lossy visual feedback that cannot resolve dimensional errors. We present CADSmith, a multi-agent pipeline that generates CadQuery code from natural language. It then undergoes an iterative refinement process through two nested correction loops: an inner loop that resolves execution errors and an outer loop grounded in programmatic geometric validation. The outer loop combines exa