figure-evaluation

Evaluate scientific figures against academic standards using a VLM-as-a-judge protocol.

Updated Jun 4, 2026
One-click install
npx skills add https://github.com/zc6600/aura --skill figure-evaluation
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: figure-evaluation
Source: https://github.com/zc6600/aura/tree/main/skills/figure-evaluation
Command: npx skills add https://github.com/zc6600/aura --skill figure-evaluation

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This skill addresses the subjectivity and inconsistency in evaluating scientific illustrations, ensuring that AI-generated figures meet rigorous academic standards before submission.

Core Features & Use Cases

  • Multi-Dimensional Scoring: Evaluates figures based on content fidelity, visual design, and communication effectiveness.
  • Blind Pairwise Comparison: Uses a judge persona to perform A/B testing on generated results to determine the superior visual output.
  • Use Case: Researchers can use this to automatically critique draft figures against ground truth data to ensure logical topology and professional aesthetics are maintained.

Quick Start

Use the figure-evaluation skill to critique the uploaded image against the provided research paper text.

Frequently Asked Questions about figure-evaluation

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

FAQPage Schema
How do I evaluate scientific figures for publication quality?

To evaluate scientific figures for publication quality, use a VLM-as-a-judge protocol that applies multi-dimensional scoring to content fidelity, visual design, and communication effectiveness against rigorous academic standards.

Can I use VLM to critique AI-generated illustrations against research paper text?

Yes, you can use a VLM judge persona to critique uploaded AI-generated illustrations against provided research paper text, ensuring logical topology and professional aesthetics are maintained for academic submission.

What is blind pairwise comparison in visual verification?

Blind pairwise comparison is an A/B testing mechanism where a judge persona evaluates generated visual outputs to determine the superior illustration, reducing subjectivity in academic visual verification workflows.

Does the figure evaluation skill require a subagent persona?

Yes, the figure evaluation process requires a subagent persona to perform multi-dimensional scoring and blind pairwise comparisons based on defined academic standards for rigorous visual verification.

When do I need multi-dimensional scoring for academic research workflows?

You need multi-dimensional scoring for academic research workflows when automatically critiquing draft figures against ground truth data to ensure AI-generated illustrations meet professional publication standards before submission.