agentleFS
mannobeatsmannobeats/hyper-taste/skills/design-system/SKILL.md

design-system

Skill1 starsChanged 4 days agoMITView on GitHub

One small design system per product, in the project's own stack (CSS, Tailwind, SwiftUI, Compose, React Native) - tokens for color, type, space, radius and motion, a set of parts in every state, a spacing grid, and one page that says which token to use when. Use when starting a product's look, when screens drift apart or values are hard-coded, when the user wants a brand or accent color, custom components, or the look of an app they name.

193 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use setup-engine, team-level and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • CoralGameSkill

    setup-engine

    Configure the project's game engine and version. Pins the engine in CLAUDE.md, detects knowledge gaps, and populates engine reference docs via WebSearch when the version is beyond the LLM's training data.

    Also in 13 projects that use this

  • CoralGameSkill

    team-level

    Orchestrate level design team: level-designer + narrative-director + world-builder + art-director + systems-designer + qa-tester for complete area/level creation.

    Also in 12 projects that use this

  • lilangMaxSkill

    create-stories

    Break a single epic into implementable story files. Reads the epic, its GDD, governing ADRs, and control manifest. Each story embeds its GDD requirement TR-ID, ADR guidance, acceptance criteria, story type, and test evidence path. Run after /create-epics for each epic.

    Also in 11 projects that use this

  • lilangMaxSkill

    story-done

    End-of-story completion review. Reads the story file, verifies each acceptance criterion against the implementation, checks for GDD/ADR deviations, prompts code review, updates story status to Complete, and surfaces the next ready story from the sprint.

    Also in 11 projects that use this

  • +2 more from CoralGame
  • +6 more from lilangMax

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 176 edited versions we read keep these.

  • 7

    Distinctive

    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:

    New project: write the tokens before the first screen.

    line 17
  2. Distinctive:

    The look of an app the user names: look at its screens (screenshots from the user, a browser, or computer use), measure sizes, spacing, …

    line 18
  3. Distinctive:

    Native apps: follow the person's system accent by default.

    line 38
  4. Distinctive:

    Keep a gallery: one page or screen that shows every part in every state, light and dark.

    line 67
  5. Distinctive:

    Use its tokens; never raw colors, radii or durations."

    line 74
  6. Distinctive:

    No raw color, radius, spacing or duration outside the token files (search the code for them).

    line 83
  7. Distinctive:

    Every part a screen needs exists with all its states, and the gallery shows them in light and dark.

    line 84

“Common” is measured: kept by at least 6 in 10 of the 176 edited versions of this file we could read. “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 system

The outcome: every screen is built from a short list of named tokens and parts, so one change in one place changes the whole product, and any agent can review a screen against one page. The system serves the direction from ui-director or mac-design; it does not replace it.

1. Read before you write

  • Existing project: list every raw value in the code (colors, opacities, radii, paddings, widths, durations, curves, custom controls). Most products have 3–5 of each where 1–2 are meant. Extend what exists in its own format; never add a second system beside it.
  • New project: write the tokens before the first screen.
  • The look of an app the user names: look at its screens (screenshots from the user, a browser, or computer use), measure sizes, spacing, type, radii and colors from them, and write those numbers down as tokens. Take the craft, never its logo, icons, text or name.

2. Tokens in three layers

  1. Palette: raw values, named by hue and step (gray-50, blue-600). Only the token file reads it.
  2. Meaning: text, text-2, text-3, surface, surface-2, line, accent, on-accent, accent-text, danger, each with a light and a dark value (and a stronger one for Increase Contrast on native platforms). Every part reads only this layer.
  3. Scales: type roles (title, body, label, meta, named by role, never by size), space, radius, control heights, motion intents (hover, press, appear, dismiss).

A rebrand changes the palette. A dark mode fix changes the meaning layer. Neither touches a part.

Rules that hold across products:

  • Accent text uses a deep shade in light mode and a light shade in dark mode. Text on an accent fill uses on-accent, never a fixed white (white fails on yellow and green).
  • Native apps follow the person's system accent by default; a brand color shows only where it means something (the icon, an empty state, one hero element).
  • Dark is its own palette with more distance between tones, not an inversion.
  • Never copy a platform's semantic colors' values; they change per release. Reference them.
  • One family for UI text, a display face for headings only with a reason, and every custom face has a bold weight.

3. The grid

  • A 4px (or 4pt) scale: 4, 8, 12, 16, 20, 24, 32, and the section spacing. Raw numbers are 0, 1, 2 or a multiple of 4.
  • Related items 8 apart, groups 20–24, sections 32 or more. One page margin.
  • Each column has one leading edge. Indent on purpose, by a step of the scale.
  • Let layout containers align things (grid, form layouts, baseline alignment, tabular numbers), not manual offsets. Offsets are for motion only.
  • Nested corners are concentric: inner radius = outer radius minus the padding.
  • Long text truncates with its full value in a tooltip. Nothing clips.

4. Parts

Build a part only when the platform's own control won't do. Each part:

  • Is a real control: a styled button, toggle or field from the platform, never a shape with a click handler (it loses the keyboard, focus and screen readers).
  • Has every state: rest, hover, pressed, selected, focused, disabled, in light and dark, and each state can be forced so it can be seen.
  • Reads only tokens. One icon set at one size and stroke, named through the system so the whole set can change in one place.

Keep a gallery: one page or screen that shows every part in every state, light and dark. Review a new part there before it goes into a screen.

5. The page that says when

DESIGN-SYSTEM.md (under 100 lines): one section per token group and part family, saying which token to use when. Then one line in the project's AGENTS.md: "Design rules: DESIGN-SYSTEM.md. Use its tokens; never raw colors, radii or durations."

6. Migrate

Existing screens move to tokens one screen per commit. Each screen looks the same or better after, never worse: look at it before and after.

Done when

  • No raw color, radius, spacing or duration outside the token files (search the code for them).
  • Every part a screen needs exists with all its states, and the gallery shows them in light and dark.
  • DESIGN-SYSTEM.md covers every token group and the project's AGENTS.md points to it.

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.