Anthropic Visual Style

Use this skill when generating any visual output that should look Anthropic-branded: plots, charts, diagrams, slides, or web pages.

Quick Start (matplotlib — primary use case)

from anthro_colors import use_anthropic_defaults use_anthropic_defaults()

import matplotlib.pyplot as plt fig, ax = plt.subplots()

All plots now use Anthropic style automatically

This loads ~/.config/matplotlib/stylelib/anthropic.mplstyle which sets:

• White background (#FFFFFF )

• PRETTY_CYCLE colors (see references/colors.md )

• No top/right spines, clean typography

• 300 DPI saves with tight bbox

Domain-Specific References

Load the relevant reference for your output type:

Domain Reference When

matplotlib references/matplotlib.md

Python plots, charts, figures

Colors references/colors.md

Color palette lookup for any domain

HTML/CSS references/web-css.md

Web pages, HTML artifacts

TikZ references/tikz.md

LaTeX diagrams for papers

Key Colors (quick reference)

Name Hex Use

DARK_ORANGE (BOOK_CLOTH) #B86046

Primary accent, first in cycle

GREY #656565

Secondary, neutral elements

DARK_BLUE #40668C

Tertiary accent

SLATE (GREY_950) #141413

Text, axes

IVORY (GREY_050) #FAF9F5

Light backgrounds (brand)

CLAY #D97757

Warm accent

SKY #6A9BCC

Cool accent

OLIVE #788C5D

Nature/green accent

Full palette with all 9 hue ramps (orange through red, 100-900 each) in references/colors.md .

Ground Truth

All color values come from lib/plotting/anthro_colors.py — that file is the single source of truth. If a hex code here conflicts with that file, the file wins.