← Back to Skills Marketplace
scientific-drawing-skill-1-0-0
by
davidzhao30
· GitHub ↗
· v1.0.0
· MIT-0
85
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install scientific-drawing-skill-1-0-0
Description
基于 AutoFigure-Edit 的科研级科学插图生成与编辑系统，能够从长篇方法描述自动生成完全可编辑的矢量图（SVG），支持参考图风格迁移和浏览器内交互式编辑
Usage Guidance
Before enabling this skill: 1) Review the referenced GitHub repository code (ResearAI/AutoFigure-Edit) yourself — the skill will ask you to clone and run that code. 2) Be cautious providing GEMINI_API_KEY or any other API tokens; prefer a least-privilege or disposable key and confirm which provider the installation actually uses. 3) The skill may download large models (SAM3) requiring HF_TOKEN and substantial disk/RAM/GPU resources — plan for storage and bandwidth. 4) The editor/server defaults to 0.0.0.0 (network-exposed); change it to bind to localhost (127.0.0.1) or put it behind an authenticated reverse proxy if you don't want public access. 5) Consider running the skill in an isolated environment/container, inspect the repo and requirements (un-pinned versions), and verify network endpoints before granting secrets or allowing automatic install.
Capability Analysis
Type: OpenClaw Skill
Name: scientific-drawing-skill-1-0-0
Version: 1.0.0
The scientific-drawing skill bundle is a well-documented tool for generating and editing scientific illustrations using AI models (Gemini, SAM3). The requested permissions, including API keys (GEMINI_API_KEY, HF_TOKEN) and Python dependencies (torch, flask, opencv), are entirely consistent with its stated purpose of image processing and vector graphics generation. While it includes a local web server for interactive editing, this is a core feature of the tool, and the bundle contains no evidence of malicious intent, data exfiltration, or prompt injection attacks.
Capability Assessment
Purpose & Capability
The skill claims to use an LLM backend and image models to produce SVGs. Requiring python3 and a language-model API key (GEMINI_API_KEY) is reasonable. However, the SKILL.md says Gemini or OpenRouter are supported and that HF_TOKEN is needed for SAM3 downloads, yet registry metadata marks only GEMINI_API_KEY as required/primary — a minor inconsistency in declared vs. actual optional/required credentials.
Instruction Scope
Instructions tell the agent to clone an external GitHub repo, install many ML packages and models, run Python scripts that read/write arbitrary filesystem paths, and launch a local editor server. The server defaults to binding 0.0.0.0 (exposes the editor to the network). There are also hardcoded example paths (/home/davidzhao/...) and open-ended batch operations that could process many files. These behaviors go beyond simple prompt-to-image steps and have operational security implications.
Install Mechanism
This is an instruction-only skill but the metadata includes pip install instructions for many heavy ML packages (torch, torchvision, opencv, etc.). Using GitHub as the source (ResearAI/AutoFigure-Edit) and PyTorch official index is expected for ML tooling, but there are no pinned package versions and the workflow downloads large models (SAM3) from Hugging Face — both increase supply-chain/compatibility risk.
Credentials
The registry requires a single primary env var (GEMINI_API_KEY), which matches LLM usage. SKILL.md, however, lists additional optional credentials (OPENROUTER_API_KEY, BIANXIE_API_KEY, HF_TOKEN) and model path env vars; HF_TOKEN is effectively required to fetch SAM3 models. The mismatch between declared required env and other tokens needed for full functionality is confusing and could lead users to provide more credentials than they expect.
Persistence & Privilege
The skill does not request always:true or any elevated platform persistence. It runs locally, and autonomous invocation is allowed by default (normal). Nothing indicates it modifies other skills or global agent config.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install scientific-drawing-skill-1-0-0 - After installation, invoke the skill by name or use
/scientific-drawing-skill-1-0-0 - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Editable SVG Output: No longer generates static PNGs; instead, produces fully editable SVG files.
Structured Components: Every icon, module, and connector is an independent, editable object.
Vector Precision: Infinitely scalable without quality loss, perfectly suited to academic publishing requirements.
Intelligent Style Transfer:
Upload a style reference image, and the AI automatically learns the color palette, typography, and icon style.
Maintains visual consistency with a laboratory or journal’s design language.
Built-in Interactive Editor:
Immediately enters a visual editing interface after generation.
Supports drag-and-drop operations to adjust layouts, modify annotations, and replace icons.
What-you-see-is-what-you-get, with undo/redo support.
Five-Stage Generation Pipeline:
Style-conditioned image generation → SAM3 segmentation and structural indexing → asset extraction → SVG template generation and refinement → asset injection
Metadata
Frequently Asked Questions
What is scientific-drawing-skill-1-0-0?
基于 AutoFigure-Edit 的科研级科学插图生成与编辑系统，能够从长篇方法描述自动生成完全可编辑的矢量图（SVG），支持参考图风格迁移和浏览器内交互式编辑. It is an AI Agent Skill for Claude Code / OpenClaw, with 85 downloads so far.
How do I install scientific-drawing-skill-1-0-0?
Run "/install scientific-drawing-skill-1-0-0" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is scientific-drawing-skill-1-0-0 free?
Yes, scientific-drawing-skill-1-0-0 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does scientific-drawing-skill-1-0-0 support?
scientific-drawing-skill-1-0-0 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created scientific-drawing-skill-1-0-0?
It is built and maintained by davidzhao30 (@davidzhao30); the current version is v1.0.0.
More Skills