eyfel/mcp-server-solidworks

This MCP server integrates with SolidWorks API and structuring it into Claude-compatible context streams.

What it solves

SolidPilot provides a bridge between AI models and SolidWorks, allowing AI to automate CAD (Computer-Aided Design) tasks. It solves the problem of "token explosion"—where exposing thousands of raw API methods to an AI would be too costly and complex—by introducing a high-level abstraction layer. Instead of calling raw API methods, the AI reasons in terms of "design intent" (e.g., "put a hole in the top face"), which is then converted into a CAD-neutral intermediate representation (IR) and compiled into concrete SolidWorks operations.

How it works

The system uses a four-layer architecture:

  1. Planner/Intent Layer: An AI model uses a schema to turn user intent into a CAD-neutral Feature Graph IR.
  2. Compiler Layer: A deterministic (non-AI) compiler lowers this IR into ordered tool calls and resolves geometric references (like "top face") against the live model state.
  3. Execution Layer: A C# .NET 4.8 application that serves as the only layer interacting with the SolidWorks COM API.
  4. Adapter Layer: A Python-based MCP (Model Context Protocol) server that acts as the boundary between the AI client (e.g., Claude, OpenAI, or local LLMs) and the rest of the system.

It also supports a "reverse" workflow where it can analyze 2D technical drawings (.DXF or .DWG files) and reconstruct the 3D part in SolidWorks based on the extracted dimensions and geometry.

Who it’s for

Mechanical engineers, CAD designers, and developers building AI agents for industrial design and manufacturing who want to automate the creation, modification, and reverse-engineering of SolidWorks parts and assemblies.

Highlights

  • CAD-Neutral IR: Uses a Feature Graph IR to separate design intent from software-specific API calls.
  • 2D to 3D Reconstruction: Capable of reading .DXF/.DWG files to automatically rebuild 3D parts, with verification against title-block weight and topology.
  • Broad AI Compatibility: Built on the open MCP standard, making it compatible with any MCP-capable AI client.
  • Comprehensive Toolset: Includes 47 tools for operations like sketching, extruding, sheet metal work, assembly mating, and detailed drawing generation.
  • Objective Verification: Includes tools to compare parts and assemblies to ensure the reconstructed model matches the original in volume, area, and topology.

Related

  • Project
  • Project
  • Project
  • Project
  • Project