agentleFS
mantleCurvemantleCurve/claude-skills/skills/unexpected-accent-palette/SKILL.md

unexpected-accent-palette

Skill0 starsChanged 10 days agoMITView on GitHub

Use when choosing or fixing a colour palette for a website, app, brand, slide deck or poster; when a two-colour brand or design feels flat, generic, forgettable, basic or "fine but boring"; when picking a CTA, highlight or accent colour; or when writing colour tokens, a theme, or light/dark colour variables.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run scripts from this skill's folder: python3 scripts/accent.py <command> ...

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Unexpected accent palette

Two base colours set the mood. One unexpected, high-chroma accent, used sparingly on the thing to click or the focal point, makes the design memorable. You don't need more colours, you need one that surprises.

Why: the isolation (von Restorff) effect makes the one item that differs get noticed and remembered, and pre-attentive pop-out lets a lone distant hue be found at a glance. Both need the accent to stay rare. Every extra use or extra hue weakens it.

Roles: 2 + 1

RoleWhat it isTypical area
Mood (BASE1)dominant base, usually the page backgroundmost of the view
Support (BASE2)second base: brand surfaces, secondary buttons, links, chart seriessome
Accentone hue far from both bases (its tints and shades are allowed)~3–10% of a viewport

Neutrals derived from the bases and status colours (error, warning, success) aren't extra hues. Keep them quieter than the accent.

Workflow

Run scripts from this skill's folder: python3 scripts/accent.py <command> ... (Python 3.9+, stdlib only).

  1. Identify the base pair from the brief, brand, CSS or screenshot. BASE1 (first argument) is the Mood/page background. Keep the user's colours. Change a base only where a needed use fails WCAG, and then only its lightness, minimally, saying so. tokens does this and records it.
  2. Analyze: analyze BASE1 BASE2 [ACCENT] [--extra HEX ...]. It prints OKLCH, hue distances, a contrast matrix and PASS/NOTE/WARN/FAIL checks. Pass the design's other colours to --extra to catch a second accent hue. Exit 1 means a FAIL.
  3. No accent yet, or it FAILs: suggest BASE1 BASE2 [--mode light|dark] [--count 6] gives deterministic, ranked candidates.
  4. Pick one with reasons: hue distance, temperature flip, chroma, contrast, status-colour clearance, brand fit. Candidates sit at the gamut edge. Softening chroma 10–20% is fine; run analyze again after.
  5. Tokens: tokens BASE1 BASE2 ACCENT --mode light and --mode dark (--format css|tailwind|json, --selector '[data-theme=dark]'). Use its hexes as they are. They cover hover, pressed, soft, disabled, on-accent, focus and status colours, with every text pair verified.
  6. Place the accent using the rules below.
  7. Verify by looking. preview BASE1 BASE2 ACCENT -o <project>/palette-preview.html renders the base pair alone vs. with the accent. On a screenshot of the real page, coverage shot.png ACCENT (needs ffmpeg) warns at 0% or over 15% and fails over 30%.
  8. Report in the format below.

Never compute contrast by hand or write "≈4.6:1, please confirm in a checker". Paste the script's output.

Quick reference

CheckTargetWhere
Hue distance (OKLCH)≥ 90° from each chromatic base. 45–90° passes only with a warm/cool flip. Under 45° FAILs. Near-neutral bases are ignoredanalyze hue-distance
TemperatureCool pair → warm accent, or the reversetemperature
Chroma≥ 0.12, ideally above the bases. Under 0.07 isn't an accentchroma
Accent fill vs page≥ 3:1 for icon-only controls, focus rings and chart marks. A labelled button only needs its label to passaccent-fill
Label on accent≥ 4.5:1, colour chosen by the scripton-accent token
Accent as textOnly where it passes: 4.5:1 for body text, 3:1 for ≥ 24px (or ≥ 18.66px bold)accent-as-text
Status colours≥ 30° from the accent hue, quieter, always with icon + textsemantic, tokens
Area~3–10% of the view. Over 15% is sprawlcoverage

Placement

Accent goes on: the primary CTA (one per view), one highlighted word or badge, a focal glow, one chart series or key data point, the active state or current selection, and the focus ring.

Never on: body text, large backgrounds or sections, several competing CTAs, decoration everywhere (every icon, border or heading), status messages, or anything that signals by colour alone.

States, dark mode, data viz, colour vision and WCAG: placement-and-accessibility.md. The nine reel palettes, measured: examples.md.

Red flags: stop and fix

  • The "accent" is the base hue again, e.g. #2F6BFF beside #09318D (1° apart). analyze FAILs it.
  • The CTA is in a base colour while the accent decorates something else.
  • You replaced the user's base pair. The pair sets the mood; the fix is the accent.
  • There's a fourth hue (a violet glow, teal icons) beside the accent. --extra FAILs it.
  • Contrast is estimated or "to be verified".
  • The accent is on two competing buttons, on paragraphs, or used as a section background.
  • The error red can't be told apart from a red or pink accent.

Report format

  1. Palette table: | Role | Hex | OKLCH | Area | Use | for Mood, Support, Accent, the tokens you used and the status colours.
  2. Why this accent: hue distance, temperature and chroma, with numbers from analyze.
  3. Where the accent goes: list every element, plus the coverage result if measured.
  4. Contrast: paste the script's lines verbatim (analyze checks and the tokens contrast block, for each mode made).
  5. Deliberately not accent: e.g. secondary buttons, links, headings, icons, errors.
  6. Changes to the user's colours: "none", or the token note (e.g. "accent lightness −0.070 → #DF7400 for 3:1 on #F8FAFF in light mode").

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.