Summary

This project offers an open-source harness designed to generate CAD models using AI coding agents. It targets developers and engineers seeking to automate 3D model creation within their workflows, providing capabilities for AI-driven design, export to multiple standard formats, and a local inspection viewer. The harness aims to streamline the iterative design process by enabling precise, code-based modifications and local execution.

How It Works

The harness facilitates text-to-CAD generation by acting as an intermediary between AI coding agents (like Codex and Claude Code) and programmatic CAD libraries, primarily build123d and OpenCascade (OCP). Its core innovation lies in a @cad[...] reference system, allowing AI agents to precisely target and modify specific geometric elements in iterative design loops. The workflow involves describing the desired model, enabling the agent to edit source files, regenerating target outputs (e.g., STEP, STL), and inspecting the results via a local viewer.

Quick Start & Requirements

• Installation: Clone the repository. Set up a Python 3.11 virtual environment and install Python dependencies (pip install -r requirements-cad.txt). Navigate to the viewer directory and install Node.js dependencies (npm install).
• Prerequisites: Python 3.11+, Node.js.
• Running: Launch the local CAD Explorer viewer using npm run dev. The viewer is accessible at http://localhost:4178.
• Links: GitHub Repo, build123d, OpenCascade, X @soft_servo.

Highlighted Details

• AI-Powered Generation: Integrates with coding agents for text-to-CAD model creation.
• Extensive Export Options: Supports STEP, STL, DXF, GLB, topology data, and URDF robot descriptions.
• Local Development: Features a local CAD Explorer viewer and runs entirely locally without backend hosting.
• Precise Referencing: Utilizes a @cad[...] reference system for iterative, geometry-aware edits by AI agents.
• Bundled Skills: Includes integrated skills for CAD operations and URDF robot description generation.

Maintenance & Community

The project appears actively maintained, with features like AI integration and a local viewer. Updates and project-related information are linked via the author's X account (@soft_servo). Dedicated community channels (e.g., Discord, Slack) are not specified in the README.

Licensing & Compatibility

The project is licensed under the MIT License. This permissive license allows for broad commercial use and integration into closed-source applications without significant restrictions.

Limitations & Caveats

This project functions as a "harness," integrating existing AI models and CAD libraries; its effectiveness is dependent on the performance and capabilities of these underlying components. Specific benchmarks or known limitations regarding the AI generation process are not detailed. The modular structure, involving "bundled skills" and separate repositories for standalone skills, may introduce complexity for advanced configurations.