solidworks-automation-skill
Reliable AI Skill + MCP toolkit for agent-driven desktop CAD automation.
Install / Use
claude mcp add wzyn20051216 -- npx -y github:wzyn20051216/solidworks-automation-skillIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AutomationSupported Platforms
Our assessment of solidworks-automation-skill
solidworks-automation-skill scores 82/100 on our quality scale, 2334th of 2,870 Automation skills we index.
Its MCP Server is 7.4 KB long, well organised into 17 sections with 4 code examples: a thorough specification that gives an agent plenty to work with.
With 1,070 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated today, so solidworks-automation-skill is actively maintained.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 95/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
solidworks-automation-skill compared with similar skills
All 4 of these similar skills score higher than solidworks-automation-skill; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| solidworks-automation-skill (this skill)by wzyn20051216 | 82 | 1.1k | today | MCP Server |
| Agent-Reachby Panniantong | 100 | 91.8k | 20d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.5k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.2k | today | CLAUDE.md |
| Scraplingby D4Vinci | 100 | 85.8k | 1d ago | MCP Server |
Frequently asked questions
- How do I install solidworks-automation-skill?
- Run
claude mcp add wzyn20051216 -- npx -y github:wzyn20051216/solidworks-automation-skill. The install tabs above show the steps for each supported agent. - Which AI agents does solidworks-automation-skill work with?
- It is written for Claude Code, Claude Desktop and OpenAI Codex, as a MCP Server file. Other agents that read the same format can often use it too.
- Is solidworks-automation-skill safe to use?
- It is MIT-licensed and scores 95/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is solidworks-automation-skill still maintained?
- The repository was last updated today, so solidworks-automation-skill is actively maintained.
Skill content
View source on GitHubSolidWorks Automation Skill
AI Skill + MCP toolkit for reliable desktop CAD automation.
让 Codex、Claude、Cursor 等 Agent 通过自然语言调用真实 CAD 软件,并对执行结果进行验证与追踪。
Tool success is not treated as task success.
工具调用成功,不代表任务真的完成;需要 Review 的任务会经过结果验证后再交付。
User → Agent (Codex / Claude / Cursor) → Skill / MCP → Execution Core → CAD → Artifacts + Evidence
AI 负责理解意图、查证 API、规划步骤;Execution Core 保证执行过程可追踪、可验证、失败可恢复。
See It In Action
<p align="center"> <img src="assets/showcase/solidworks-engineering-drawing-demo.gif" alt="代码驱动 SolidWorks 生成复杂测试件与 GB/T 工程图" width="900"> </p>一个真实流程:自然语言需求 → Agent 调用 Skill → SolidWorks 生成原生 SLDPRT + SLDDRW → 导出 PDF → 自动复核尺寸与证据。
| 案例 | 能力 | 状态 |
|---|---|---|
| M6×1 真实螺纹孔 | Metric Tap Thread,重建后回读 | verified |
| CNC 多圆角/倒角安装座 | 语义选边、可变半径、face/full-round/setback | stable |
| 标准渐开线直齿轮 | 模数/齿数/压力角驱动实体 | pilot |
Quick Start
三种入口任选其一即可,不需要同时启动。
Option 1 — Skill(最简单)
npx github:wzyn20051216/solidworks-automation-skill
# 或
claude skill add https://github.com/wzyn20051216/solidworks-automation-skill
Option 2 — MCP
npm install -g @smithery/cli
smithery mcp add wzyn20051216/solidworks-automation-skill --client codex --config '{}'
Option 3 — CAD Studio 桌面版
从 GitHub Releases 下载 Windows 安装包。
Choose one entry point. You do not need to run all three.
| 入口 | 适合谁 | 需要 GUI | 典型用法 | |---|---|---|---| | Skill | 支持 skill 导入的 Agent 客户端 | 否 | 自然语言驱动 SolidWorks | | MCP | Codex / Claude / Cursor / Windsurf 等 MCP 客户端 | 否 | Agent 自动 Tool Calling | | CAD Studio | 需要图形界面管理项目/任务/预览/交付 | 是 | 桌面端提交任务 |
三者是平级入口,共用同一套能力清单(capabilities.yaml)、脚本与证据系统。安装后可用 python scripts/cad_doctor.py 做环境诊断;详细安装与 CLI 见 docs/CAD_STUDIO_USER_MANUAL.md。
Why This Project
LLM 会规划,但常见的 AI CAD 自动化有几个真实问题:
- LLM 会规划,但不会真正操作桌面 CAD 软件
- Tool 返回
success不代表任务真的完成 - CAD API 版本、环境、Backend 差异复杂
- 固定 Tool 集合覆盖不完所有工程需求
- Agent 执行过程通常不可追踪、不可审计
本项目的解法:
| 问题 | 方案 | |---|---| | 自然语言 → 真实操作 | Skill / MCP 接入真实 COM 与后端 | | Tool success ≠ Task success | Verification-first completion | | Backend 失败 | Backend Router + Recovery Decision | | 能力缺失 | Capability Gap → 交还 Agent 查证 | | 执行不可追踪 | 结构化 Trace | | 是否可靠 | Golden Workflow Eval |
Architecture
User / Agent
↓
Skill / MCP
↓
Execution Core(Capability · Verification · Recovery Decision · Trace)
↓
Backend Router
↓
Python COM / C# / OCCT
↓
CAD Software
↓
Artifacts + Evidence
-
Root Skill = 通用能力、规则与执行基础设施(连接、建模、审查、MCP)。
-
Subskills = 专项领域工作流(工程图、螺纹孔、CNC 圆角、VibeCAD、AutoCAD),由 Root Skill 按需调用,不需要用户手动启动。
-
Execution Core = 内部可靠执行层:统一状态、结果验证、失败恢复决策与执行追踪。它是内部实现,不是独立服务、不需要用户启动。
-
Capability Gap = 找不到能力时结构化返回
CAPABILITY_GAP,交还 Agent 查证,而不是「做不了」:Existing Tools → Tool Composition → Backend Fallback → Capability Gap → Agent 查 references / 官方 API / SDK → 最小实现 → VerificationExecution Core 自身不会自动联网执行未知代码。
详细:docs/architecture/execution-core.md
Reliability
V2 附带 37 个 deterministic reliability scenarios,覆盖正常执行、故障注入、后端回退、能力缺口与策略门禁。
| Metric | Current baseline | |---|---| | Nominal workflow | 11/11 first-pass | | Injected false-completion cases | all detected | | Escaped false completion | 0 in current deterministic benchmark | | Regression suite | 740 passed / 1 skipped / 0 failed |
这是 deterministic engineering benchmark(Fake Handler/Reviewer + 故障注入), 不是生产 SLA 或真实用户成功率。详见
docs/architecture/v2-evaluation.md。
Capabilities & Status
| Category | 你能做什么 | 状态 |
|---|---|---|
| Modeling | 零件建模、草图/拉伸/旋转、齿轮、钣金、焊件 | verified + pilot |
| Assembly | 装配、Mate、干涉检查、爆炸视图 | verified |
| Drawing | GB/T 工程图、尺寸链、孔表、BOM | verified + pilot |
| Export | STEP / STL / IGES / PDF / DXF / DWG | verified |
| Motion | 运动算例、旋转马达、结果门禁 | pilot |
| FEA | CalculiX 线性/非线性求解编排与审计 | pilot |
| DFM / Routing | 制造复核、路由复核 | pilot |
| Surface / Mold | 受限 Loft/Sweep/Thicken、模具计划 | pilot |
| AutoCAD | DWG/DXF 二维绘图、图层/标注/预览 | verified / pilot / blocked |
| Open-format | OCCT 无头写入 STEP/BREP/STL/OBJ/GLB 等 | verified |
状态等级(capabilities.yaml 定义):
| Status | 含义 |
|---|---|
| verified | 真机验证完成,可按已知范围使用 |
| stable | 长期回归覆盖 |
| pilot | 已实现但仍需人工复核 |
| blocked | 当前环境或能力不允许自动交付 |
真机基线:SolidWorks 2024 / 2026、AutoCAD 2024。未验证能力不会被包装成无人值守交付。
- 完整能力矩阵(每个能力的精确等级/版本/限制)→
capabilities.yaml - 领域工作流 →
subskills/·SUBSKILLS.md
No SolidWorks Installed?
没有 SolidWorks 时,部分开放格式任务仍可完成:
STEP · IGES · BREP · STL · OBJ · GLB · DXF · SVG · PDF · PNG
(通过 OCCT/OCP 隔离进程真实写入。)
但原生 SLDPRT / SLDASM / SLDDRW 仍需要合法 SolidWorks 环境。
Examples
examples/ 目录包含从基础零件到装配、工程图、Motion、FEA 的可运行示例。
Documentation
| Topic | Link |
|---|---|
| MCP Setup | mcp-server/README.md |
| CAD Studio | docs/CAD_STUDIO_USER_MANUAL.md |
| Architecture | docs/architecture/execution-core.md |
| Reliability Eval | docs/architecture/v2-evaluation.md |
| Capability Matrix | capabilities.yaml |
| Subskills | SUBSKILLS.md |
| API / Backend | references/ |
| Troubleshooting | references/troubleshooting.md |
Roadmap & Contributing
Roadmap:更广的真机验证 · 更可靠的能力路由 · 更好的 CAD Studio 可观测性 · 更多 verified 子技能 · 社区贡献工作流。
Contributing:新增能力时,先查 capabilities.yaml,优先复用现有 backend,新增 tests + evidence,不要把未验证 capability 标为 verified。详见 CONTRIBUTING.md。
License & Community
MIT · GitHub Issues · Star History
English
SolidWorks Automation Skill is an AI Skill + MCP toolkit for reliable desktop CAD automation. Codex, Claude, Cursor, and other agents drive real SolidWorks / CAD workflows through natural language, with verification, deterministic recovery decisions, and execution tracing. Three peer entry points — Skill, MCP, and CAD Studio — share one capability registry (capabilities.yaml). Choose one; no extra runtime service is required. See Quick Start above.
<sub>关注抖音 @balance. · 嵌入式开发、SolidWorks 自动化和 AI 辅助工程实践持续更新</sub>
Related Skills
Agent-Reach
91.8kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
74.5kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.2kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Scrapling
85.8k🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
