Can I modify a figure after it is generated?

Yes, the 'Critique' phase allows Claude to view the generated image and perform iterative revisions based on your feedback.

How does it ensure figures are publication-quality?

The skill follows a dedicated 'Stylist' agent phase that applies NeurIPS 2025 style guidelines for colors, typography, and layout.

Is the Gemini API free for this tool?

Yes, you can obtain a free API key from Google AI Studio which currently supports up to 500 images per day without a credit card.

What are the requirements to use PaperBanana?

You need Python, the 'google-genai' and 'Pillow' packages, and a free Google Gemini API key to enable image generation.

Can it generate both diagrams and statistical charts?

Yes, it creates architectural diagrams via Gemini's native image generation and generates Matplotlib code for precise statistical plots.

PaperBanana Scientific Illustration Generator

Name: PaperBanana Scientific Illustration Generator
Author: varunr89

byvarunr89

0•

数据科学与机器学习

Generates publication-ready scientific diagrams and statistical plots for academic papers using a multi-agent pipeline.

PaperBanana is a specialized tool for researchers that automates the creation of high-quality figures for conference submissions and journals. It orchestrates a sophisticated pipeline where Claude acts as a planner, stylist, and critic to transform methodology text and captions into professional visuals. By adhering to NeurIPS-style guidelines and leveraging the Gemini API for native image generation or Matplotlib for data plots, PaperBanana ensures your research architecture and results are communicated with professional-grade clarity and aesthetic precision.

主要功能

010 GitHub stars

02Automated adherence to NeurIPS-style guidelines for diagrams and plots

03Reference-driven generation using the PaperBananaBench dataset

04Iterative feedback loop for high-fidelity visual and textual accuracy

05Multi-agent pipeline for planning, styling, and critiquing scientific visuals

06Support for both architectural flowcharts and data-driven statistical charts

使用场景

01Generating publication-quality statistical plots directly from data descriptions

02Creating 'Figure 1' overview diagrams for machine learning conference papers

03Visualizing complex system architectures and methodological pipelines for research

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add varunr89/claude-marketplace paper-banana

For use in Claude.ai and ChatGPT

主要功能

010 GitHub stars

02Automated adherence to NeurIPS-style guidelines for diagrams and plots

03Reference-driven generation using the PaperBananaBench dataset

04Iterative feedback loop for high-fidelity visual and textual accuracy

05Multi-agent pipeline for planning, styling, and critiquing scientific visuals

06Support for both architectural flowcharts and data-driven statistical charts

使用场景

01Generating publication-quality statistical plots directly from data descriptions

02Creating 'Figure 1' overview diagrams for machine learning conference papers

03Visualizing complex system architectures and methodological pipelines for research

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add varunr89/claude-marketplace paper-banana

For use in Claude.ai and ChatGPT