Text to CAD Model Generator

The Problem

Turning natural-language descriptions into usable CAD models usually stalls on engineering details: dimensions, features, constraints, and valid geometry. Hand-writing CadQuery scripts often fails on fillets, booleans, and face or edge selection. This skill turns text-to-model into an executable, debuggable, deliverable workflow.

How It Works

• Environment setup: Prefer an existing Python, conda, or virtual environment with CadQuery; install only if none is found.
• Requirement parsing: Extract geometry type, dimensions, holes, fillets, patterns, and assembly relationships; missing units default to mm.
• Script generation: Produce a self-contained Python script with parameters grouped at the top. Prefer .faces() and .edges() for selective edits, and apply fillets or chamfers after boolean cuts.
• Execution and delivery: Run the script to generate STL and STEP files, apply bounded auto-debugging on failure, and verify file existence, bounding-box dimensions, and requested features before delivery.

Boundaries

It is useful for generating mechanical parts, housings, flanges, bearing seats, and bolt-hole patterns, or for pre-print geometry checks. It is not a substitute for formal engineering design: stress, assembly tolerance, and material manufacturability still require human review. Complex thin walls or multi-body assemblies still need manual decomposition. If the description is purely aesthetic and lacks functional constraints, the result will favor common engineering interpretations.