1. Sourcing & naming
New palettes come from three places: our own design work, community submissions through
the upload page, and gaps we identify in the library's coverage —
for example, adding depth to categories with fewer than a certain number of quality
entries. Nothing is scraped from other tools.
Every palette carries a specific, human-readable name ("Santorini Sunset", not
"Palette 1731") chosen to describe what the combination actually evokes. A palette only
earns a place in the library if it can be described in a sentence: where it would live,
what it communicates, and why the four colors belong together.
Rejections are common: near-duplicates of existing schemes, combinations with no
discernible use case, and "technically harmonious but ugly" ramps all get cut. Volume
is never the goal — the library grows only when a new
entry is genuinely worth a designer's time.
2. Color-theory scoring
Each palette is scored against the five classical harmony rules — complementary,
analogous, triadic, split-complementary and tetradic — by measuring the hue positions
of its colors on the color wheel. Palettes that don't sit close enough to one of these
relationships are flagged and either re-tuned or rejected.
The harmony rule a palette follows is displayed on its detail page, along with a hue
wheel showing the exact relationships, so you can judge the structure yourself rather
than taking our word for it.
3. Accessibility testing (WCAG 2.1)
Every palette is tested for WCAG 2.1 contrast compliance before publication. We compute
relative luminance for each color and score the background/text pairing that the
palette's design roles imply, then publish the grade — AAA, AA, or Fail — on the
palette's page alongside the exact contrast ratio.
Every palette page also shows a full contrast matrix: each color tested
as text on every other color as background, with the safe use cases for each ratio. We
would rather tell you a palette fails for body text than let it ship as a "dark mode"
scheme that isn't readable.
4. Uniqueness enforcement
The library's build pipeline rejects any new palette whose exact color ramp already
exists under a different name. Slugs are unique across all 2,700+ entries, and every
category page (by mood, dominant
color, or use case) is checked for thin coverage
before each release so the library stays balanced instead of lopsided.
5. Ongoing maintenance
The library is a living thing, not a one-time dump. Each release cycle we:
- Re-check contrast data and fix any values that no longer match our scoring pipeline;
- Deduplicate slugs and color ramps flagged by the build's uniqueness checks;
- Review category pages for thin or overlapping coverage;
- Refresh examples, pairings and usage guidance on existing palette pages.
Tool pages evolve too — the AI palette generator and
Palette Studio are re-tested with every release of
the scoring engine so their output stays consistent with the library's standards.
6. Corrections & feedback
If you find a wrong hex value, a mislabeled mood, or an accessibility claim that doesn't
hold up, tell us via the contact page and we'll fix it —
corrections to factual data (hex codes, contrast ratios) take priority and usually ship
within a release cycle.
For anything else about how the library works — licensing (all palettes are free for
commercial use; individual colors can't be copyrighted), bulk access, or attribution —
the FAQ covers the common cases.