콘텐츠로 이동

illustration-designer — playbook

Design and spec an illustration system, OR a single illustration within an existing system. Output is a markdown spec a designer / illustrator can produce against, OR direct SVG output for simple cases.

When to use

  • "We need an illustration system for our SaaS landing page."
  • "Spec the empty state illustrations for our product."
  • "Design a mascot for our Korean fintech app."
  • "Should we use Lottie or SVG for the hero?"
  • "Optimize this SVG."

Inputs (ask if missing)

  1. Scope — full system, OR single illustration.
  2. Surface — empty state / hero / onboarding / spot / mascot.
  3. Brand voice — 3-5 adjectives (e.g., calm, gentle, money-warm; or sharp, futuristic, controlled).
  4. Existing system? — if yes, follow its style. If no, this skill defines one.
  5. Color tokens — which brand tokens are available.
  6. Format constraint — SVG / Lottie / PNG / WebP / mixed.
  7. Korean market? — affects mascot vs minimal-geometric default.
  8. Production budget — solo designer, agency, AI-assisted, none yet.

Steps

1. Classify the work

Task Approach
Define a new system Start with illustration-systems.md, then apply the surface-specific illustration references
Add to existing system Follow style guide; don't redefine
Single spot illustration spot-illustrations.md
Hero illustration hero-illustrations.md
Mascot work mascot-design.md
SVG technical / optimization svg-optimization.md

2. If defining a system — set the style

From illustration-systems.md:

  • Geometric vs organic — pick one.
  • Line weight rule — stroke-only / fill-only / mixed; one stroke width.
  • Color treatment — monochromatic / brand-palette / limited-extended.
  • Geometry rules — grid, corner radius, stroke caps.
  • Perspective — flat / isometric / front-view.
  • Voice — 3-5 adjectives.

State each choice with reasoning.

3. Pick the format

Format Use
SVG Default; vector, themeable
Lottie Animated illustration
PNG / WebP Heavy shading, painterly, photo-realistic
AVIF Highest compression for raster

For most: SVG. Lottie when motion is the point. Raster only when SVG can't represent the visual.

4. Per-illustration spec

For each illustration:

Name: <kebab-case, prefixed by category>
       e.g., illo-empty-search.svg, illo-hero-onboarding-01.svg
Surface: <empty-state | error-state | onboarding-step-2 | hero-section-3>
Subject: <one sentence — what's depicted>
Voice: <encouraging | calm | celebratory | apologetic>
Size: <display dimensions>
Format: <SVG | Lottie | PNG>
Tokens used: <list of CSS variables / brand colors>
Animation? <static | subtle loop | one-shot>
Accessibility: <decorative | meaningful + alt text>
Production hours: <estimate>

5. Apply Korean conventions (if applicable)

6. Optimize the SVG

For each shipped SVG:

  • Run SVGO with config from svg-optimization.md
  • Replace brand color hex with currentColor or CSS variable
  • Verify viewBox preserved
  • Verify no embedded raster
  • Verify no editor metadata
  • Verify size targets: < 1kB icon, < 10kB spot, < 50kB hero
  • Test at 1×, 2×, 3×
  • Test in light + dark modes

7. Accessibility

  • Decorative: aria-hidden="true" on illustration.
  • Meaningful: role="img" + aria-label.
  • Animated: respect prefers-reduced-motion.

8. Output

Use this structure:

# Illustration spec: <system or piece name>

> Scope: <system | single piece>
> Style: <flat geometric | soft 3D | hand-drawn | mascot-driven | editorial | generative>
> Voice: <3-5 adjectives>
> Format: <SVG | Lottie | mixed>

## Style rules
<line weight, color treatment, geometry, perspective>

## Color tokens
<list of brand tokens permitted>

## Asset list
| Name | Surface | Size | Format | Tokens | Anim? |

## Production checklist
<checklist per illustration>

## Don't
<2-3 specific misuses>

Source files this skill reads

Verification phase (run before declaring done)

  • If defining a system: are style + voice + color rules all explicit?
  • Is the format chosen consciously (SVG default; Lottie / PNG only with reason)?
  • Are color tokens listed (no raw hex in the spec body)?
  • Are illustrations sized per category (1kB icon / 10kB spot / 50kB hero)?
  • Is accessibility specified (decorative vs meaningful + alt text)?
  • If animated: is reduced-motion fallback specified?
  • If Korean B2C: are mascot conventions or Toss-style restraint applied?
  • If shipping SVG: is SVGO + currentColor + viewBox checklist passed?
  • Does the "Don't" section catch 2-3 specific misuses?

Done when

  • One markdown spec.
  • Style + voice + format choices explicit and reasoned.
  • Color tokens listed.
  • Asset list with sizes, formats, accessibility.
  • Production checklist.
  • "Don't" section.
  • Verification passes.