agentleFS
sergeyizmailovsergeyizmailov/knowledge-delta-skills/skills/frontend/design-stack-picker/SKILL.md

design-stack-picker

Skill3 starsChanged 49 days agoMITView on GitHub

Use when building or restyling frontend UI and choosing building blocks — icon sets, fonts, component/block libraries, accessible primitives, imagery sources, color systems, motion, shadows, spacing, and image optimization. For landing pages, ecommerce, dashboards, web apps, emails, and single components.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Add a library only when there is no good local solution.

    line 19
  2. Distinctive:

    No dependency for: a single icon when an icon system exists.

    line 37
  3. Distinctive:

    Add one when it removes real implementation risk: complex accessible widgets, broad consistent icon coverage, image optimization at scale, …

    line 39
  4. Distinctive:

    Never hand-draw generic SVG icons — use the icon set (supplied brand logos excepted).

    line 55
  5. Distinctive:

    Verify in a browser from phone to wide desktop: no horizontal overflow, key images load at natural dimensions.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

design-stack-picker

A selection layer, not a mandate to replace the project's design system.

Decision order

Audit first: framework and rendering model, existing icons/fonts/tokens/components/image pipeline/breakpoints, current dependencies, visual style. Then:

  1. Reuse the existing project system if it works.
  2. Extend it with the smallest compatible building block.
  3. Add a library only when there is no good local solution.
  4. Hand-craft only for brand assets, tiny static pieces, or where a dependency outweighs the implementation.

Defaults

Starting points, not requirements.

AxisDefault
IconsIconify + Solar; Simple Icons for brand marks
FontsUnbounded display + Onest body via Fontsource
ComponentsProject components first; HyperUI/Preline for static sections; shadcn/Radix for React primitives
ColorSemantic CSS variables; one dominant brand color + one accent
MotionCSS-first restrained reveal; Motion for React; GSAP only for complex timelines
ImageryReal product/user/stock/CMS assets; AVIF/WebP where practical

Dependency budget

No dependency for: a single icon when an icon system exists; one static section existing primitives can handle; a basic disclosure/dropdown in Astro or plain HTML where scoped JS is smaller and verifiable; a visual effect existing CSS covers; anything duplicating a current dependency.

Add one when it removes real implementation risk: complex accessible widgets, broad consistent icon coverage, image optimization at scale, multi-screen design systems.

Context routing

ContextBias
Astro/staticLocal .astro components, scoped CSS, progressive JS, astro:assets; avoid React-only primitives unless already used
React/NextCurrent UI layer first; shadcn/Radix for missing complex primitives
Admin/dashboardDense scannable forms/tables/filters, minimal decoration
EcommerceReal product photos, neutral placeholders, stable card dimensions with object-fit: contain; price/stock/spec/action hierarchy over decorative labels; reachable category nav, autocomplete, dismissible mobile filters; test long names, missing images, sale/stock states, uneven counts
MarketingStronger typography, imagery/video, section blocks, restrained motion
PrototypeSmallest working choice; no full design system unless it will continue

Hard rules

  • One family per axis: one icon set, one illustration style, one color system, one type pairing.
  • Never hand-draw generic SVG icons — use the icon set (supplied brand logos excepted).
  • Choose and load type intentionally; never ship default fonts on brand-facing UI.
  • Tokens for colors, spacing, radii, shadows, type, and states — no scattered inline magic values.
  • Accessible primitives for dialogs, comboboxes, tabs, menus, tooltips, switches — with real keyboard/focus behavior.
  • Respect prefers-reduced-motion, visible focus, AA contrast.
  • Optimize important images, set dimensions, eager-load only the LCP image.
  • Verify in a browser from phone to wide desktop: no horizontal overflow, key images load at natural dimensions.

Where a hard rule conflicts with a mature existing convention, keep the convention and improve incrementally.

Resources

Load only what the task needs, never both by default.

  • resources.md — the catalog: libraries, fonts, icon sets, imagery, components, motion, color systems. §12 first for vague "make it look better" tasks, to fix concrete references before touching UI.
  • patterns.md — implementation snippets: tokens, reset, icons, fonts, cards, image markup, motion, CSS.

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.