Scientific Graphical Abstract Skill

Generate editable, publication-quality SVG graphical abstracts for scientific papers using AI models with customizable charts, diagrams, and data-driven visu...

Security Scan

The skill mostly matches its stated purpose (SVG graphical abstracts) but shows several inconsistencies: it references optional API keys and external install steps that are not declared in metadata, and README contains remote install commands (curl | sh) and an external GitHub repo reference — you should verify these before installing or providing credentials.

Purpose & Capability

The name/description (generate editable SVG graphical abstracts) aligns with the included code and examples. However, the README and SKILL.md advertise integrations with Anthropic/ OpenAI/DeepSeek APIs while the skill metadata declares no required environment variables — that's a mild mismatch (API use is optional in code, but metadata doesn't document these optional creds).

Instruction Scope

SKILL.md and README instruct the agent/user to load local CSV/JSON files and optionally use AI model APIs. That scope is appropriate for the feature set, but the runtime instructions assume availability of API keys and local files while metadata omits those env declarations. The README also suggests running external install steps and placing files under ~/.claude/skills, which will write to user disk; the instructions grant broad discretion to call external APIs if keys are provided.

Install Mechanism

The registry lists no install spec, but README suggests installation from a personal GitHub repo and provides a remote install pipeline: an npx clone URL and a curl | sh command to install 'uv' (https://astral.sh/uv/install.sh). Remote install scripts and cloning unverified repos increase risk; the skill lacks an official, vetted install spec in the registry.

Credentials

The code (Config._load_api_keys) reads ANTHROPIC_API_KEY, OPENAI_API_KEY, and DEEPSEEK_API_KEY from the environment and README documents these optional vars, but the skill metadata declares no required env vars. Requesting API keys for multiple providers is understandable for multimodel support, but the metadata omission is an inconsistency and increases the chance users provide credentials without clear justification. Provide keys only if you trust the code and owner.

Persistence & Privilege

The skill does not request always:true, does not declare system-wide config paths, and README describes installing into the user's skills directory (~/.claude/skills). Nothing in the package requests elevated or persistent privileges beyond normal installation.

What to consider before installing

Key points to consider before installing: - Verify the source repository and author (README points to a personal GitHub). Only install from a repo/account you trust. - The README suggests running a remote installer (curl https://astral.sh/uv/install.sh | sh). Avoid piping unknown scripts to sh; review the script first or use an alternative installation method. - The code will read ANTHROPIC_API_KEY / OPENAI_API_KEY / DEEPSEEK_API_KEY from your environment if present. The skill metadata did not declare required env vars — do not supply API keys unless you trust the code and repository. If you must provide keys, create least-privilege credentials and monitor usage. - The included Python script appears to depend on listed requirements; consider running it in an isolated virtual environment or sandbox before granting any credentials. - There is a minor code quality issue (math is used in several methods but not imported in the visible snippet) — this suggests limited review/testing; review the full code for other bugs or hidden network calls. - If you want to proceed, review the full repository contents locally (search for network calls, external endpoints, and code that reads arbitrary files outside the skill folder), or ask the publisher to add clear requires.env metadata and an official install spec.

Scientific Graphical Abstract Generator

AI-Powered Scientific Visualization Tool - Generate editable graphical abstracts for research papers

Description

Generate professional graphical abstracts for scientific papers using multiple AI models. This skill transforms research descriptions and data into publication-quality, editable SVG visualizations.

Features

• Multiple AI Models: Support for Claude, GPT-4o, DeepSeek, and other multimodal models
• Editable SVG Output: Generate vector graphics that can be edited in Inkscape, Illustrator, or any SVG editor
• Multiple Chart Types: Bar charts, line charts, pie charts, scatter plots, flowcharts, diagrams
• Data-Driven Visuals: Automatically create charts from CSV/JSON data
• Customizable Styles: Adjust colors, fonts, layouts to match journal requirements
• Publication Ready: High-quality output suitable for scientific journals

Usage

/graphical-abstract Generate a graphical abstract showing the research workflow for a CRISPR gene editing study
/graphical-abstract Create a bar chart comparing the performance of three different machine learning models from this data: [data]
/graphical-abstract Design a flowchart illustrating the mechanism of action of the proposed drug
/graphical-abstract Generate a line chart showing temperature changes over time from the following CSV data

Examples

Generate Research Workflow Diagram

/graphical-abstract Create a graphical abstract showing the workflow: Sample preparation → RNA extraction → Sequencing → Data analysis → Results visualization. Use a clean, professional style with blue color scheme.

Create Data Visualization

/graphical-abstract Generate a bar chart with the following data:
Model A: 85% accuracy
Model B: 92% accuracy
Model C: 78% accuracy

Title: Model Performance Comparison
Y-axis: Accuracy (%)
Color scheme: Professional blue gradient

Generate Mechanism Diagram

/graphical-abstract Design a schematic diagram showing how the proposed inhibitor binds to the active site of the enzyme, blocking substrate access. Include labels for key components.

Command Reference

generate

Generate a graphical abstract.

Option	Description
--prompt	Description of what to visualize
--data	Data file (CSV/JSON) for charts
--type	Chart type: bar, line, pie, scatter, flowchart, diagram
--model	AI model: claude, gpt4o, deepseek (default: claude)
--style	Style: minimal, professional, colorful, journal
--output	Output SVG file path
--width	Canvas width (default: 800)
--height	Canvas height (default: 600)

template

Use predefined templates for common visualizations.

Option	Description
--type	Template type: workflow, mechanism, comparison, timeline
--prompt	Specific requirements

Supported Models

Claude (Recommended)

• Best for: Complex diagrams, scientific illustrations
• Vision capabilities: Excellent
• API: Anthropic Claude API

GPT-4o

• Best for: Charts, data visualization
• Vision capabilities: Very good
• API: OpenAI API

DeepSeek

• Best for: Technical diagrams, cost-effective
• Vision capabilities: Good
• API: DeepSeek API

Output Format

All outputs are in SVG format which offers:

• Editable: Open in Inkscape, Adobe Illustrator, or any text editor
• Scalable: Infinite resolution without quality loss
• Web-ready: Can be embedded directly in websites
• Publication quality: Meets most journal requirements

Example SVG structure:

<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 800 600">
  <!-- Editable elements with clear structure -->
  <g id="chart-area">...</g>
  <g id="labels">...</g>
  <g id="legend">...</g>
</svg>

Notes

• For complex visualizations, provide detailed descriptions
• Data files should be in CSV or JSON format
• SVG files can be edited after generation
• Different models may produce different styles
• Journal-specific requirements can be specified in the prompt

Related Skills

• sci-data-extractor - Extract data from scientific papers
• semanticscholar-search-skill - Search academic papers
• pubmed-search-skill - Search biomedical literature