How does the quality review system work?

Gemini 3 Pro evaluates each generated image on five key criteria: Scientific Accuracy, Clarity, Label Quality, Layout/Composition, and Professional Appearance, assigning a total score out of 10.

Does it regenerate images every time?

No, it uses 'smart iteration.' It only regenerates the diagram if the Gemini 3 Pro quality score falls below the specific threshold for your selected document type (e.g., 8.5 for journals or 6.5 for presentations).

Can I customize the visual style of the diagrams?

Yes, you can specify layout (e.g., 'vertical flow'), quantitative details (e.g., 'n=500'), and style preferences in your natural language prompt, and the AI will adapt the output accordingly.

Where are the generated diagrams stored?

All diagrams are automatically stored in the figures/ subfolder of your project, accompanied by a JSON review log containing the quality scores and critiques.

What types of diagrams can this skill generate?

The skill can generate a wide range of technical visualizations including neural network architectures, CONSORT/PRISMA flowcharts, biological pathways, circuit diagrams, and system block diagrams.

Scientific Schematics AI

Name: Scientific Schematics AI
Author: plurigrid

byplurigrid

•

Data Science & ML

Generates publication-quality scientific diagrams and neural network architectures using AI-driven iterative refinement and quality scoring.

This skill automates the creation of complex scientific visualizations, ranging from biological pathways to deep learning architectures, by leveraging Nano Banana Pro AI for generation and Gemini 3 Pro for rigorous quality review. It implements a smart iteration workflow that evaluates diagrams against document-specific standards—such as Nature-grade journal requirements or conference poster thresholds—ensuring professional-grade output while minimizing API usage. Ideal for researchers and engineers, it eliminates the need for manual drawing tools by transforming natural language descriptions into polished, versioned schematics ready for publication.

Key Features

01Automated quality review and scoring by Gemini 3 Pro against 5 scientific criteria

02Smart iterative refinement based on document-type thresholds (Journal, Conference, Poster, etc.)

03AI-powered generation of neural network architectures, flowcharts, and biological pathways

04Standardized output following scientific best practices including colorblind-friendly palettes

058 GitHub stars

06Detailed versioned review logs with quality scores, critiques, and early-stop metrics

Use Cases

01Visualizing complex Transformer, CNN, or system architectures for technical papers

02Generating biological signaling pathways and molecular interactions for academic posters

03Creating CONSORT or PRISMA flowcharts for clinical trial and medical publications

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi scientific-schematics

For use in Claude.ai and ChatGPT

Download Skill