Linear UI Icon Set

Have AI hand you four style-consistent 48×48 minimal linear UI icons that drop straight into your interface.

When to use it

Add icons to a UI:

I'm building a web dashboard or an app screen and need a set of feature icons with a shared visual language — I want AI to give me four candidates in one go.

Describe the product area:

I describe the domain in one sentence (payments / cloud storage / messaging / calendar) and let AI pick the right graphic metaphor for each icon, so I don't have to spell out shapes one by one.

Extend an icon set:

My interface already uses a soft-pastel + 2px rounded-stroke icon style; new icons must extend that language rather than shift to a different look.

Limit the colors:

I want only two colors — say, blue and green — or only warm tones. The whole set should stay within the palette I dictate.

Set size and stroke:

I've already committed to a 48×48 square backing. Any other canvas size is off-scope.

Not for:

• Brand logos or brand marks — use a dedicated brand-design tool
• Full-page UI mocks or posters — route to ui-ux-pro-max
• Architecture or flow diagrams — route to fireworks-tech-graph
• 3D, skeuomorphic, cartoon, doodle, or pixel styles

What you get

Always four standalone square icons — never a merged poster, never a full-page UI mock. That's the whole point.

• Canvas: four 48×48px canvases with square (non-rounded) backings; the subject sits centered at ~24–30px with generous margins
• Stroke: uniform 2px linear stroke on the final canvas, rounded caps and rounded joins, matching visual weight across all four
• Palette: each icon pairs one soft pastel background with a deeper matching stroke color — the default pool is #FFF3E8 orange / #F1EAFE purple / #EAF2FF blue / #EAF8F2 green
• Graphics: four distinct metaphors or structures, but stroke, proportion, negative space, and corner handling stay identical
• Never does: add text, digits, letters, watermarks, or logos; apply gradients, shadows, highlights, or textures; render rounded backings or transparent backgrounds

Prerequisites & boundaries

Prerequisites:

Any prompt-to-image model (Midjourney, DALL·E, Nano-banana, and so on). This skill produces the prompt — it does not render pixels itself.

Adjacent skills:

Need	Route to
Full-page UI mock / poster / landing visual	ui-ux-pro-max
Architecture / flow / data diagram	fireworks-tech-graph
Brand logo / trademark / brand mark	Not covered — use a brand-design tool

Out of scope:

• Brand logos, trademarks, or brand-identity marks
• Full-page UI, posters, or promotional visuals with copy
• 3D, skeuomorphic, cartoon, doodle, or pixel styles
• Transparent backgrounds, rounded backings, or elaborate containers

Subtle boundaries:

• Asking for a count other than 4 (3, 6, 8) → not triggered; four is a hard constraint
• Asking for a canvas other than 48×48 (128×128, 32×32) → not triggered; the size is part of the constraint
• Asking for two saturations of the same hue → still triggered, as long as one color per icon holds