# Figure

> Generates complete, runnable visualization code for research figures. Produces matplotlib/seaborn Python scripts for quantitative data and Mermaid diagrams for conceptual relationships, workflows, and taxonomies. Use when the user asks for charts, graphs, visualizations, figures, or diagrams. Produces code only — does not render images directly.

- Skill: `richard-kim-79/figure` (Agent Skill)
- Install (CLI): `npx skillmds@latest add richard-kim-79/figure`
- Raw SKILL.md: https://api.skillmd.com/api/skills/richard-kim-79/figure/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- License: MIT
- Author: richard-kim-79 (https://skillmd.com/u/richard-kim-79)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/richard-kim-79/figure

---


# Figure Generator

Generate publication-quality visualization code for research figures.

## Figure Types

| Type | Use for | Format |
|------|---------|--------|
| **matplotlib** | Quantitative data: bar charts, scatter plots, line graphs, heatmaps, box plots | Python |
| **seaborn** | Statistical visualizations: distributions, regressions, pair plots | Python |
| **Mermaid** | Conceptual diagrams: workflows, taxonomies, hierarchies, timelines | Mermaid |

## Decision guide

- Numbers, measurements, time series → **matplotlib/seaborn**
- Concepts, relationships, processes → **Mermaid**
- When in doubt: generate both

## Output Format

````markdown
## Figure 1: [Title]

**Caption:** [Full figure caption as it would appear in a paper — what is shown and the key takeaway]

**Description:** [What this figure shows and why it matters for the research]

**Type:** Python (matplotlib/seaborn)

```python
import matplotlib.pyplot as plt
import numpy as np

# [Complete, runnable code with realistic placeholder data]
plt.tight_layout()
plt.savefig('figure1.png', dpi=300)
plt.show()
```
````

## Code requirements

- **Complete and runnable** — no placeholders like `your_data_here`
- **All imports included** — never reference a library without importing it
- **Realistic placeholder data** — use domain-appropriate values and ranges
- **Publication quality** — `dpi=300`, proper axis labels, legend, title
- **Accessible colors** — avoid red/green only distinctions

## Validation (always run after generating Python code)

Before presenting code to the user, validate syntax with Python's AST parser:

```bash
python -c "import ast; ast.parse(open('figure1.py').read()); print('✅ Syntax OK')"
```

If the check fails, fix the syntax error and re-validate before showing the result.

**For inline code blocks**, validate with:
```python
import ast
code = """
# paste generated code here
"""
try:
    ast.parse(code)
    print("✅ Syntax OK")
except SyntaxError as e:
    print(f"❌ Syntax error: {e}")
```

> Note: AST validation checks syntax only — it does not catch runtime errors (e.g. wrong data shapes). Always include comments explaining how to adapt placeholder data to real data.

## How to run

**Python figures:**
```bash
pip install matplotlib seaborn numpy
python figure1.py
```

**Mermaid diagrams:**
- Paste into [mermaid.live](https://mermaid.live)
- Or use any Markdown renderer that supports Mermaid (Obsidian, GitHub, etc.)