PaperBanana — Academic Illustration Generator

What it does

PaperBanana is a CLI-first multi-agent pipeline (Retriever → Planner → Stylist → Visualizer → Critic) for producing academic diagrams, statistical plots, and presentation slides. It orchestrates VLMs and image providers to generate, critique, and refine visual outputs suitable for publications and presentations.

When to use it

Use when you need AI-assisted generation of methodology figures, data visualizations from CSV/JSON, single or batch slide creation, or comparative evaluation between generated images and human references. It's intended for research authors, presenters, and visualization workflows where iterative critique improves output quality.

What's included

• Scripts: CLI-based commands (generate, plot, slide, slide-batch, evaluate, data, setup) described in the skill documentation; no embedded scripts flagged in fetch (has_scripts=false) but rich command examples are provided.
• References: provider selection guidance, style presets, and argument parsing tables are included in the skill body.
• Instructions: installation and configuration notes, provider keys (.env), recommended flags, error handling and recovery, and best practices for safe auto-runs.

Compatible agents

Designed for Claude Code / agents that can run CLI commands and orchestrate external providers (Gemini, Anthropic, OpenAI, Bedrock, OpenRouter). The skill assumes network access and provider API keys in environment configuration.