Introducing vibe-cading, an AI native parametric model generator powered by CadQuery

11 views
Skip to first unread message

Marcus Chen

unread,
Jun 27, 2026, 6:33:46 PM (9 days ago) Jun 27
to CadQuery
Hi all,

I wanted to share vibe-cading (open source, PyPI, AGPLv3), a project built on top of CadQuery.

The foundation is a library of reusable parametric generator classes — not one-off scripts, but composable component factories with a consistent interface:

- Mechanical primitives — screws, hex nuts, heat-set inserts, bearings, gears (ISO-validated spur gears via SpurGear.from_iso(...)), hinges, joints, standoffs, enclosures.
- Cutter pattern — components expose a .to_cutter(profile=...) method, so a screw or axle can bore its own clearance/counterbore into a host body via boolean ops, instead of you hand-modeling the negative.
- Lego Technic + RC — Technic axles, beams, and cross-axle hole cutters, plus RC-to-Technic adapters (motor mounts, ESC holders, axle-to-bearing sleeves).
- Per-printer tolerance profiles — fits are driven by a profile you calibrate once, so the same generator produces FDM-ready parts across different machines instead of hard-coded clearances.

Install: pip install vibe_cading

The library is designed to be driven by an LLM. An optional MCP server exposes the engine to any MCP client (Claude Desktop, Cursor, etc.) with tools to list and inspect the generator classes, pull design context, and compile a model to STEP/STL — and the same interface lets an agent author a new CadQuery generator when nothing in the library fits. Compilation runs locally on your machine. The models in the library are themselves authored with AI assistance (Claude Code / Antigravity), and the agent workflow is shipped in the repo.

It's early and built solo, with a deliberately narrow focus on maker / mechatronics and FDM-printable mechanisms. I'd genuinely value feedback from this group — especially on the generator interface and tolerance model. Issues and contributions welcome.

Source: https://github.com/fa-mc/vibe-cading
PyPI: https://pypi.org/project/vibe-cading/

Regards,
Marcus
Reply all
Reply to author
Forward
0 new messages