Graph Design

Graph Design Skill

Produces publication-quality matplotlib figures using an iterative render-critique loop. Output goes to /mnt/user-data/outputs/ as both PNG (1000 DPI) and SVG.

🔄 The Optimization Loop (Required Workflow)

Never submit a plot without running this loop. Do not skip the critique step.

1. Draft — Implement the plot using the configurations below.
2. Render — Generate the image file.
3. Critique — Use the view tool to visually inspect the output against the Quality Checklist.
   • Ask: "Is the font serif? Are top/right spines removed? Do markers have white edges? Is the aspect ratio correct?"
4. Refine — Fix any discrepancies found in the critique.
5. Verify — Re-render and re-critique until all checklist items pass.

When NOT to Use This Skill

• Quick throwaway/exploratory plots with no quality requirements
• Plotly, Seaborn, Bokeh, or any non-matplotlib backend
• Purely statistical output (tables, descriptive stats) with no figure needed

Figure Dimensions

Project defaults (override per request if needed):

Configuration Reference

Typography

plt.rcParams.update({
    'font.family': 'serif',       # Critical — never leave as default sans-serif
    'font.size': 10,
    'axes.labelsize': 11,
    'axes.titlesize': 12,
    'xtick.labelsize': 9,
    'ytick.labelsize': 9,
})

Color Palette (Paul Tol — colorblind-safe)

COLORS = {
    'primary':     '#88CCEE',
    'secondary':   '#CC6677',
    'tertiary':    '#DDCC77',
    'quaternary':  '#6699CC',
    'quinary':     '#888888',
    'accent':      '#EE7733',
    'grid':        '#CCCCCC',
}

For sequential data use: viridis, plasma, magma, or inferno.

Marker Specifications

Z-Order Hierarchy

Export Settings

OUTPUT_DIR = '/mnt/user-data/outputs/'   # Always use this path

def save_fig(fig, name):
    fig.savefig(f'{OUTPUT_DIR}{name}.png', dpi=1000, bbox_inches='tight', facecolor='white')
    fig.savefig(f'{OUTPUT_DIR}{name}.svg', dpi=72,   bbox_inches='tight', facecolor='white')

Clean Aesthetics (L-Frame Pattern)

Always remove top and right spines. Always render grid behind data.

fig, ax = plt.subplots(figsize=(FIG_WIDTH, FIG_HEIGHT))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.grid(True, which='major', linestyle='-', alpha=0.5, zorder=1)

Complete Starter Template

Copy and adapt this for every new plot:

import os
import numpy as np
import matplotlib.pyplot as plt

# --- Output ---
OUTPUT_DIR = '/mnt/user-data/outputs/'
os.makedirs(OUTPUT_DIR, exist_ok=True)

def save_fig(fig, name):
    fig.savefig(f'{OUTPUT_DIR}{name}.png', dpi=1000, bbox_inches='tight', facecolor='white')
    fig.savefig(f'{OUTPUT_DIR}{name}.svg', dpi=72,   bbox_inches='tight', facecolor='white')
    print(f"Saved: {name}.png + {name}.svg → {OUTPUT_DIR}")

# --- Config ---
FIG_WIDTH, ASPECT_RATIO = 6.0, 1.3
FIG_HEIGHT = FIG_WIDTH / ASPECT_RATIO

COLORS = {
    'primary': '#88CCEE', 'secondary': '#CC6677', 'tertiary': '#DDCC77',
    'quaternary': '#6699CC', 'quinary': '#888888', 'accent': '#EE7733', 'grid': '#CCCCCC',
}

plt.rcParams.update({
    'font.family': 'serif', 'font.size': 10,
    'axes.labelsize': 11, 'axes.titlesize': 12,
    'xtick.labelsize': 9, 'ytick.labelsize': 9,
})

# --- Data ---
x = np.logspace(1, 6, 100)
y = 1 / x + 0.01

# --- Plot ---
fig, ax = plt.subplots(figsize=(FIG_WIDTH, FIG_HEIGHT))
ax.set_xscale('log')
ax.set_yscale('log')

ax.plot(x, y, color=COLORS['primary'], linewidth=2, zorder=3)
ax.scatter(x[::10], y[::10],
           color=COLORS['secondary'], s=10**2,
           edgecolors='#FFFFFF', linewidths=1.5, zorder=3)

ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.grid(True, which='major', linestyle='-', alpha=0.5, zorder=1)

ax.set_xlabel('Frequency (Hz)')
ax.set_ylabel('Amplitude')
ax.set_title('Result', fontweight='bold')

plt.tight_layout()
save_fig(fig, 'final_plot')

✅ Quality Checklist

Run through this during the Critique step. Do not skip.

• Font: Is font.family explicitly set to 'serif'? (Check rcParams, not just appearance)
• Spines: Top and right spines removed?
• Grid: Grid lines behind data (zorder ≤ 2)?
• Colors: From Paul Tol palette or a perceptually uniform sequential map?
• Markers: White edge (#FFFFFF), edge width 1.5, size 10?
• Aspect ratio: Does the figure match the intended column width?
• Labels: All axes labelled, font size ≥ 9pt, legible at output size?
• Export: Both PNG (1000 DPI) and SVG saved to /mnt/user-data/outputs/?