A pigment-inspired colour palette for reading and code, rooted in Moby-Dick. Warm paper on one end, deep ink on the other, with eight accent hues named after the crew of the Pequod.
- Base scale: twelve steps from Log 50 (warm paper) to Log 950 (night sky).
- Accents: eight crew members — Ahab, Starbuck, Queequeg, Pip, Ishmael, Stubb, Tashtego, Daggoo — each with a light and a dark variant and a recommended syntax role.
- Read-first: designed for long-form reading and code at length, not for glance-ability. Saturation stays low, backgrounds stay warm, accents stay in the same pigment register.
- Accessibility: every body-text pair clears WCAG-AA (4.5:1) on the reference surface; dark-mode accents clear 4.5:1 comfortably. Colour- vision-deficiency collapses are documented, not hidden — see below.
- Semantics, not decoration: each accent has a role. Using the palette should feel earned; the colour choice should tell you something.
The full narrative and design rationale live at https://tiagojct.eu/projects/pequod/. This repository is the canonical source of truth for the tokens and the built themes.
Alpha (0.1.0). Tokens are stable enough to run a website on them and to publish across every ecosystem the palette ships into:
| Surface | Where | Status |
|---|---|---|
| VS Code extension | Marketplace, Open VSX | live |
| Zed theme | themes/Pequod.zed.json |
drop-in |
| Terminal presets | themes/terminals/ |
Ghostty, Alacritty, kitty, WezTerm, tmux, Windows Terminal, iTerm2 |
| Python package | PyPI | pip install pequod |
| Tailwind plugin | npm | npm install pequod-tailwind |
| R package | CRAN | install.packages("pequod") |
| Specimen PDF | specimen/specimen.pdf |
regenerated from tokens |
Hex values may still shift by a point or two during the alpha — palette
testing continues against more code and long-form prose. Breaking
changes before 1.0 will be called out in CHANGELOG.md.
Eight hero plots in matplotlib — sequential heatmaps, grouped and horizontal bars, scatter, box plots, distributions, time series, and a specimen-style swatch grid. Four on dark, four on light:
Titles in Atkinson Hyperlegible Next SemiBold, ticks in JetBrains Mono, surfaces and series
colours pulled directly from pequod.LOG, pequod.CREW_LIGHT, and
pequod.CREW_DARK. Full gallery, code patterns, and reproduction
instructions live in examples/. Generate them
yourself with:
pip install "pequod[plot]" numpy
python examples/plots.pypequod/
├── pequod.json # the canonical palette tokens
├── themes/
│ ├── Pequod.itermcolors # iTerm2 (dark)
│ ├── Pequod-color-theme.json # VS Code (dark) — canonical
│ ├── Pequod-light-color-theme.json # VS Code (light) — canonical
│ ├── Pequod.zed.json # Zed (dark + light in one file)
│ └── terminals/ # Ghostty, Alacritty, kitty,
│ # WezTerm, tmux, Windows Terminal
├── tailwind/ # npm package — `npm install pequod-tailwind`
├── vscode/ # VS Code extension — `make vsix` builds
│ │ # the .vsix; published to the Marketplace
│ │ # as "tiagojct.pequod-color-theme"
│ ├── package.json
│ ├── icon.png
│ └── themes/ # copies of the canonical themes
├── python/ # Python package — install with
│ │ # pip install pequod
│ ├── pyproject.toml # see python/README.md for full Python docs
│ ├── src/pequod/ # palette + matplotlib helpers
│ ├── tests/
│ └── data-raw/ # generator: re-reads ../pequod.json
├── r/ # R package — install with
│ │ # remotes::install_github(
│ │ # "tiagojct/pequod", subdir = "r")
│ ├── DESCRIPTION # see r/README.md for full R docs
│ ├── R/ # palette constants, ggplot2 scales
│ ├── tests/
│ └── data-raw/ # generator: re-reads ../pequod.json
├── specimen/
│ ├── specimen.typ # single-page specimen source (Typst)
│ └── specimen.pdf # rendered output — swatches + samples
├── scripts/
│ └── cvd_check.py # Viénot–Brettel–Mollon CVD simulation + ΔE
├── Makefile # ~13 targets — see `make help`
├── README.md
├── CHANGELOG.md
├── LICENSE-CC-BY-4.0 # palette tokens and docs
└── LICENSE-MIT # theme files, scripts, R package
The themes ship as a marketplace extension under vscode/,
published as tiagojct.pequod-color-theme on both surfaces. Install
in any of three ways:
- Marketplace — open Extensions (⌘⇧X), search Pequod Palette, install. Or use the listing page. For VSCodium / Cursor / Gitpod, use the Open VSX listing instead.
.vsixfile — build withmake vsix, thencode --install-extension vscode/pequod-color-theme-0.1.0.vsix.- From source (no marketplace) — copy the
vscode/folder to~/.vscode/extensions/pequod-color-theme/. VS Code will pick it up on next launch.
Then Preferences: Color Theme → pick Pequod or Pequod Light.
Zed reads user themes directly from disk:
- Copy
themes/Pequod.zed.jsonto~/.config/zed/themes/Pequod.zed.json(create the folder if it does not exist). - Restart Zed → theme selector: toggle → pick Pequod Dark or Pequod Light.
- Open Settings → Profiles → Colors → Color Presets → Import…
- Select
themes/Pequod.itermcolors. - Apply the Pequod preset.
An iTerm2 light preset is on the roadmap.
Drop-in dark presets for the most common terminals live in
themes/terminals/:
| Terminal | File |
|---|---|
| Ghostty | Pequod.ghostty |
| Alacritty | Pequod.alacritty.toml |
| kitty | Pequod.kitty.conf |
| WezTerm | Pequod.wezterm.lua |
| tmux | Pequod.tmux.conf |
| Windows Terminal | Pequod.windowsterminal.json |
See themes/terminals/README.md for the
install path each terminal expects.
npm install pequod-tailwind// tailwind.config.js
const pequod = require("pequod-tailwind");
module.exports = {
theme: {
extend: {
colors: pequod.colors, // log + all eight crew accents
},
},
};<body class="bg-log-50 text-log-800 dark:bg-log-950 dark:text-log-100">
<h1 class="text-queequeg dark:text-queequeg-dark">Pequod</h1>
</body>Full usage in tailwind/README.md.
pip install pequod # palette + helpers
pip install "pequod[plot]" # adds matplotlib gluefrom pequod import LOG, CREW_LIGHT, palette
palette("log") # 12-step Log scale, list of hex
palette("crew", n=5) # first five crew accents
# matplotlib (with the [plot] extra)
import matplotlib.pyplot as plt
import pequod
pequod.register_cmaps()
plt.imshow(data, cmap="pequod_log")Full usage in python/README.md.
The R package is on CRAN:
install.packages("pequod")
library(pequod)
palette_pequod("log") # 12-step Log scale
palette_pequod("crew", n = 5) # first five crew accents
pequod_preview("crew") # quick base-R previewSource lives in r/; to follow the development version,
remotes::install_github("tiagojct/pequod", subdir = "r").
ggplot2 scales are provided too:
library(ggplot2)
ggplot(iris, aes(Sepal.Length, Sepal.Width, colour = Species)) +
geom_point(size = 3) +
scale_color_pequod_d(palette = "crew")Full usage in r/README.md.
pequod.json is the single source of truth. The file contains:
log— the twelve-step base scale (Log 50 → Log 950).accents— the eight crew accents, each withlight,dark,role, and a shortnoteexplaining the character and the syntax role.roles— semantic role → token mappings for light and dark modes (bg,text,text-muted,link,accent-primary, etc.).syntax— default mappings from syntax role to accent (keyword, string, comment, function, type, constant, variable, operator).
The R package, the Python package, the Typst specimen, and the CVD
test script all regenerate from pequod.json (see each one's
data-raw/ or generator). The terminal presets and the editor themes
are still produced by hand; bringing them under the same generator
contract is the next priority — see What comes next below.
Body-text contrast on the reference surfaces:
| Pair | Use | Ratio |
|---|---|---|
| Log 800 on Log 100 | light body | 10.8 : 1 |
| Log 700 on Log 50 | light link | 8.5 : 1 |
| Log 400 on Log 50 | muted / large text | 3.9 : 1 |
| Log 100 on Log 950 | dark body | 14.0 : 1 |
| Accent-light on Log 100 | UI accents | 3.0 – 9.5 : 1 |
| Accent-dark on Log 950 | dark-mode accents | 3.6 – 13.6 : 1 |
In v0.2.0-alpha every accent clears the AA-large (3 : 1) bar on its target surface — none of the bright-on-light "fill-only" cases from v0.1 remain. Light-mode body-text candidates (≥ 4.5 : 1 on Log 100): Daggoo, Queequeg, Pip, Ahab. Dark-mode body-text candidates (≥ 4.5 : 1 on Log 950): every accent except Daggoo (3.6, AA-large only).
scripts/cvd_check.py simulates each accent at 100 % severity for
protanopia, deuteranopia, and tritanopia using the Viénot–Brettel–Mollon
(1999) model, and reports pairwise ΔE*ab (CIE76, Lab D65) between
simulated accents.
In v0.2.0-alpha the worst-case pair under each simulation:
| light variants | dark variants | |
|---|---|---|
| protanopia | Ishmael ↔ Tashtego, ΔE 15.1 | Stubb ↔ Tashtego, ΔE 11.8 |
| deuteranopia | Ishmael ↔ Tashtego, ΔE 8.0 | Ishmael ↔ Tashtego, ΔE 6.8 |
| tritanopia | Pip ↔ Daggoo, ΔE 13.3 | Ahab ↔ Pip, ΔE 10.2 |
The deuteranopia floor is set by Ishmael ↔ Tashtego because green collapses to neutral grey under deutan and the two accents sit at similar L*. This is unavoidable for an 8-hue palette that includes both a green and a low-chroma grey, so it is documented rather than designed away — in practice, comments (Ishmael) and strings (Tashtego) are rarely adjacent and rarely encode meaning by colour alone.
Usage guidance: do not rely on colour alone to distinguish Ishmael from Tashtego under deuteranopia. Pair with icon, weight, or italics (the default theme already italicises comments). Every other pair clears ΔE ≥ 10 across all three simulations.
Run the check yourself:
python3 scripts/cvd_check.pyThe script has no dependencies beyond NumPy.
A one-page A4 specimen — the full Log scale, the eight crew accents
with light and dark variants, a body-text sample, and a dark code
sample with every token coloured by its crew role — lives at
specimen/specimen.pdf. Use it as a quick
reference when choosing which accent a new UI element should take,
or print it and pin it somewhere.
The PDF is generated from specimen/specimen.typ
with Typst. To regenerate after a token change:
make specimen
# equivalent to: typst compile specimen/specimen.typ specimen/specimen.pdfTypst uses the system-installed Atkinson Hyperlegible Next and JetBrains Mono. Install them from Google Fonts if they are not already present.
- Generators for the editor and terminal themes, so the
hand-maintained files in
themes/andthemes/terminals/regenerate frompequod.jsonthe same way the R, Python, and specimen targets already do. - Light presets for the terminals (currently dark only) and for iTerm2 specifically.
- Vim / Neovim colourscheme using Lush.
- Tailwind v4 plugin as a first-class plugin once Tailwind v4's
plugin API stabilises (the current package works in v4 via
@themebut isn't formally a v4 plugin yet). - Sublime Text, Helix, Emacs — lower priority.
Contributions to any of these are welcome.
- Flexoki by Steph Ango is the most direct inspiration. Pequod owes its philosophy — warm paper, cool ink, muted accents, published tokens — to Flexoki. Where Flexoki draws on earth pigments broadly, Pequod narrows the story to a ship, a century, and a text.
- Solarized by Ethan Schoonover established the modern practice of designing palettes for reading first.
- Tokyo Night and Nord are two other palettes with a disciplined colour story worth studying.
- Herman Melville, Moby-Dick; or, The Whale (1851), for the names.
- Palette tokens (
pequod.json) and documentation — Creative Commons Attribution 4.0 International. Use, adapt, ship; credit required. SeeLICENSE-CC-BY-4.0. - Theme files, scripts, and packages (
themes/*,scripts/*,tailwind/*,vscode/*,python/*,r/*,specimen/*) — MIT. SeeLICENSE-MIT.
If you use Pequod in a project or have a suggestion, I would love to hear about it: tiagojacinto@med.up.pt, or open an issue here.