freecad-mcp

freecad-mcp

MCP server that exposes parametric FreeCAD operations as LLM tools, enabling natural-language-driven CAD modeling through a headless FreeCAD environment in Docker.

Category
访问服务器

README

freecad-mcp

Hand-coded Python MCP server exposing parametric FreeCAD operations as LLM tools — the same tool-use plumbing used to give LLMs access to document stores, applied to a grounded CAD environment.

Project status

Stage: foundation complete — server under construction.

  • ✅ Repo hygiene: .gitignore, README
  • 🔜 pyproject.toml (content ready — pins MCP SDK)
  • 🔜 .omp/mcp.json placeholder (OMP harness config)
  • ✅ Step 2 — FreeCAD headless spike: proven natively (macOS) and in a Docker container
  • 🔜 Steps 3A+ — MCP server skeleton, six tools, OMP harness wiring, verification

The server code does not exist yet. This README documents the foundation and the environment decisions everything else builds on.

Architecture

  • In-process headless FreeCAD: the server imports FreeCAD/Part and runs geometry operations in its own process — no GUI, no separate CAD instance, no RPC bridge.
  • Container runtime: the server runs in a Docker image where FreeCAD is installed as a Python library alongside the interpreter. This is the only setup where import FreeCAD is guaranteed to work for any reviewer.
  • Why not native macOS? FreeCAD.app ships its Python bindings sealed inside the app bundle, compiled against its own embedded Python. Importing them from an external venv is fragile and machine-specific (a fight with rpaths and versions that breaks on every FreeCAD update). The container is reproducible by construction — and it exercises Docker, which is a stated requirement for this role.

Prerequisites

  • Docker Desktop (any platform) — server and FreeCAD live in the image.
  • Python ≥ 3.10 locally (for server development and tests).

Quick start (current stage: spike verification)

# Build the dev image (Debian bookworm, FreeCAD 0.20.2, Python 3.11 venv, MCP SDK)
docker build -t freecad-mcp .

# Verify FreeCAD imports inside the container
docker run --rm freecad-mcp app_env/bin/python -c "import FreeCAD; print(FreeCAD.Version())"

# Headless pipeline smoke: document → box → recompute → export STEP
docker run --rm freecad-mcp app_env/bin/python -c "import FreeCAD, Part; doc=FreeCAD.newDocument('t'); Part.show(Part.makeBox(10,10,10),'Box'); doc.recompute(); Part.export([doc.Objects[0]], '/tmp/t.step'); print('ok')"

Known quirks (hard-won)

  • Debian's freecad-python3 installs its bindings to /usr/lib/freecad-python3/lib — a non-standard path that Python never searches by default. The image sets ENV PYTHONPATH=/usr/lib/freecad-python3/lib to wire it up.
  • Debian's python3 strips ensurepip, so creating a venv requires the python3-venv package (python3 -m venv fails without it).
  • The macOS FreeCAD installer ships no FreeCADCmd binary — headless use on macOS goes through the app's bundled Python (/Applications/FreeCAD.app/Contents/Resources/bin/python) with PYTHONPATH pointing at Contents/Resources/lib.
  • Docker Desktop's daemon must be running before docker commands work (open -a Docker on macOS).

Roadmap

  • [x] .gitignore, README
  • [ ] pyproject.toml — Python ≥ 3.10, MCP SDK pinned (stable 1.x, spec 2025-06-18)
  • [ ] .omp/mcp.json placeholder
  • [ ] MCP server skeleton (official SDK, stdio first) — src/freecad_mcp/server.py
  • [ ] Six tools: create_document, create_primitive, execute_python, export_model, list_objects, get_object
  • [ ] OMP harness wiring + verification (/mcp list, /mcp test freecad)
  • [ ] Security hardening (sandboxed execute_python, export path guards, doc limits)
  • [ ] Verification: MCP Inspector, Claude Desktop, OMP harness; pytest; CI

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选