midhunxavier

Create Figures

Generate publication-ready figures and tables for LaTeX or DOCX formats. Creates data plots, diagrams, and tables with proper formatting.

midhunxavier 0 Updated 3mo ago

Resources

1
GitHub

Install

npx skillscat add midhunxavier/article-writing/create-figures

Install via the SkillsCat registry.

SKILL.md

Create Figures Skill

Generate publication-ready figures and tables for your paper. Acts as a Figure Generator supporting both LaTeX and DOCX formats.

Overview

This skill helps you create:

  • Data plots and charts (Python-based)
  • Architecture diagrams (TikZ)
  • Tables (LaTeX booktabs or Word)
  • Schematics and flowcharts

Commands

/create-figures <description> [--data <file>] [--format latex|docx]

Generate a figure or table.

Parameters:

  • description (required): Description of the figure to create
  • --data <file> (optional): Data source file (CSV, JSON) for data-driven plots
  • --format latex|docx (optional): Output format (default: latex)

Figure Types

Data Plots

  • Tool: Python (Matplotlib/Seaborn)
  • Use Cases: Charts, graphs, performance curves, accuracy plots
  • Input: CSV, JSON data files
  • Output: PDF images (LaTeX) or embedded images (DOCX)

Schematics and Diagrams

  • Tool: TikZ (LaTeX) or Python diagrams (DOCX)
  • Use Cases: Architecture diagrams, flowcharts, system diagrams
  • Output: TikZ code (LaTeX) or vector graphics (DOCX)

Tables

  • Tool: LaTeX + booktabs (LaTeX) or Word tables (DOCX)
  • Use Cases: Results tables, comparisons, ablation studies
  • Format: No vertical lines, proper horizontal rules

Generation Guidelines

  1. Code Choice:

    • Python for complex data plots
    • TikZ for diagrams/schematics (LaTeX)
    • Python diagrams for DOCX schematics
  2. Table Formatting:

    • Use booktabs package (LaTeX)
    • No vertical lines
    • Proper horizontal rules
  3. Vector Graphics:

    • Prefer PDF/EPS export for publication quality
    • High resolution for raster formats
  4. Consistency:

    • Match paper font sizes and styles
    • Consistent color schemes
  5. Reproducibility:

    • Save generation scripts alongside figures
    • Document data sources

Output Includes

  • Generated figure file (.pdf, .png, or TikZ code)
  • Generation script (Python or TikZ)
  • LaTeX include snippet with caption and label (LaTeX)
  • Word insertion instructions (DOCX)

Examples

# Generate accuracy comparison table
/create-figures "accuracy comparison table" --data results.csv

# Generate transformer architecture diagram
/create-figures "transformer architecture diagram with attention mechanism"

# Generate training loss curve
/create-figures "training loss curve comparison" --data experiments/logs.json

# Generate DOCX figure
/create-figures "model comparison chart" --data results.csv --format docx

Figure Placement

Generated figures are placed in:

  • figures/ directory for image files
  • tables/ directory for table files
  • TikZ code can be included directly or saved as separate files

LaTeX Integration

For LaTeX, figures include:

\begin{figure}[htbp]
    \centering
    \includegraphics[width=0.8\textwidth]{figures/figure-name.pdf}
    \caption{Figure caption}
    \label{fig:figure-name}
\end{figure}

DOCX Integration

For DOCX, figures are:

  • Inserted into Word document
  • Properly sized and captioned
  • Linked to caption numbering

Scripts Used

  • ./figure.py - Main figure generation script

Platform Compatibility

Works on all platforms (Claude Code, Codex, OpenCode, Gemini, Cursor). Command syntax may vary by platform - see platform-specific documentation in docs/.