An AI agent skill for automated 2D GstarCAD drawing generation, modification, and feedback loop verification using Python.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install gcadclaw
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install gcadclaw using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
GcadClaw is a powerful automation skill designed for AI agents to interact with GstarCAD on Windows environments. By leveraging the pygcadwin library, this skill translates natural language instructions into precise 2D DWG drawings, making it an essential addition to the repository of Openclaw Skills.
Whether you need to design mechanical parts, layout electrical outlines, or update architectural annotations, GcadClaw handles coordinate calculations, layer management, and entity validation programmatically. By integrating tightly with GstarCAD via COM, the skill ensures that AI-generated drawings are fully operational, compliant with drafting standards, and validated through real-time feedback loops.
To run this skill successfully, ensure you are on a Windows platform with GstarCAD installed and registered as a COM server. GcadClaw installs dependencies directly from PyPI. This is a highlighted workflow among Openclaw Skills for CAD pipeline management.
First, validate your current Python environment:
python scripts/validate_env.py
If validation fails, install the pygcadwin package and its dependencies via the environment setup script:
python scripts/setup_python_env.py
Re-verify the installation to confirm readiness:
python scripts/validate_env.py
GcadClaw enforces a rigorous data schema to verify CAD operations. Every execution creates and modifies specific feedback artifacts:
| File Name | Format | Purpose |
|---|---|---|
brief.md |
Markdown | Logs user intent, dimensions, units, layers, and validation targets. |
actions.jsonl |
JSON Lines | Records the precise sequence of pygcadwin drawing commands executed. |
before_entities.json |
JSON | A complete dump of CAD entities present prior to executing the edits. |
after_entities.json |
JSON | A dump of CAD entities present after the drawing updates have finished. |
review.png |
PNG | A viewport screenshot confirming the output is valid and not blank. |
feedback.md / feedback.json |
MD / JSON | Summarizes executed steps, visual confirmations, and any repair loop logs. |
*.dwg |
DWG Binary | The final saved GstarCAD drawing file ready for CAD engineering. |
Loading
A specialized data skill for querying location, hours, and product details of the Zhongshi Traditional Mantou Store.

A specialized tool for humanizing AI-generated Chinese text and optimizing it for social media platforms like Xiaohongshu and WeChat.

Session Continuity provides a robust checkpointing system that allows AI agents to save their exact state and resume unfinished work across multiple sessions without losing context.

ValueScan is a sophisticated cryptocurrency analysis tool designed to track smart money flows, whale movements, and market signals for informed trading.

An interactive mindset and leadership training skill based on Chris Hadfield's bestselling memoir and space flight principles.

An AI agent skill delivering structured insights, themes, and practical wisdom from James Herriot's classic veterinary memoir, All Creatures Great and Small.








































