paper-banana-orchestration

Orchestrate PaperBanana sub-agents to generate publication-quality academic figures.

11|3|Updated Feb 25, 2026
One-click install
npx skills add https://github.com/jiutuhky/my-super-capsule --skill paper-banana-orchestration
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: paper-banana-orchestration
Source: https://github.com/jiutuhky/my-super-capsule/tree/main/paper-banana/skills/paper-banana-orchestration
Command: npx skills add https://github.com/jiutuhky/my-super-capsule --skill paper-banana-orchestration

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Generate publication-quality figures for research papers by orchestrating the PaperBanana multi-agent pipeline, reducing manual effort in creating diagrams, method illustrations, and plots.

Core Features & Use Cases

  • Orchestrates five sub-agents (Retriever, Planner, Stylist, Visualizer, Critic) to produce diagrams and plots.
  • Supports file-based output conventions (pipeline_state.json, descriptions/, images/, code/), enabling traceability and incremental refinement.
  • Polishes figures with NeurIPS-2025 aesthetic guidelines and reads .tex/.md sources for auto-illustrations.
  • Iterative critic loop for refinement with early stopping when no changes are needed.
  • Flexible for generating diagrams (methodology, architecture, pipeline) and plots (bar charts, line charts, heatmaps).

Quick Start

Provide a Methodology section and a diagram caption, and the system will orchestrate the PaperBanana pipeline to generate and refine the figure.

Frequently Asked Questions about paper-banana-orchestration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate publication-quality academic figures from a methodology section?

You can generate publication-quality academic figures by providing a methodology section and a target caption, which triggers a multi-agent pipeline to automatically produce and iteratively refine diagrams and plots.

Can I create NeurIPS-style diagrams and plots automatically from markdown or latex sources?

Yes, you can create NeurIPS-style diagrams from .tex or .md sources by applying specific aesthetic guidelines to automatically polish the generated academic illustrations and plots.

How does the iterative refinement process work for polishing academic illustrations?

Iterative refinement for academic illustrations uses a critic loop with early stopping, evaluating generated diagrams and plots until no further visual changes are needed to reach optimal quality.

What file formats and conventions are used for storing pipeline state and generated images?

The pipeline stores state and outputs using file-based conventions including a pipeline_state.json file, alongside dedicated directories for descriptions, images, and code to ensure full traceability.

What types of diagrams and plots can I create for research papers?

For research papers, you can create methodology, architecture, and pipeline diagrams, alongside various plots including bar charts, line charts, and heatmaps tailored to academic publication standards.