Figure Generation

This implements the MatPlotAgent three-phase pipeline for generating publication-ready scientific figures. It expands your description into specs, generates matplotlib code with up to 4 retry attempts on errors, then uses visual feedback to refine the output. Handles the usual suspects: bar charts, training curves, heatmaps, ablation studies. Outputs both 300 DPI PNG and vector PDF, plus the LaTeX boilerplate to drop it in your paper. The quality checks are sensible, enforcing colorblind-friendly palettes and readable font sizes at print scale. Honestly, the VLM feedback loop is the interesting bit here. Most figure tools stop at code generation, but having it actually look at the rendered output and fix labeling or scaling issues saves the manual iteration you'd normally do.

skills.sh installs: 2.3K
GitHub stars: 459
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add lingzhi227/agent-research-skills --skill figure-generation --agent claude-code

Installs into .claude/skills of the current project.

Files

Scientific Figure Generation

Generate publication-quality figures for research papers.

Input

$0 — Description of the desired figure
$1 — (Optional) Path to data file (CSV, JSON, NPY, PKL) or results directory

Scripts

Generate figure template

python ~/.claude/skills/figure-generation/scripts/figure_template.py --type bar --output figure_script.py --name comparison
python ~/.claude/skills/figure-generation/scripts/figure_template.py --list-types

Available types: bar, training-curve, heatmap, ablation, line, scatter, radar, violin, tsne, attention

Three-Phase Pipeline (from MatPlotAgent)

Phase 1: Query Expansion

Expand the user's figure description into step-by-step coding specifications using the prompts in references/figure-prompts.md. Determine: figure type, data mapping (x/y/color/hue), style requirements, paper conventions.

SKILL.md on GitHub