All skills
hueyexe avatar

/design-systems-frontend-architecture

@2841c07

Use when building reusable frontend systems, design tokens, component contracts, responsive layouts, CSS strategy, or design-system documentation and governance.

Use this Skill: https://skilld.dev/gh/hueyexe/frontend-agent-skills/design-systems-frontend-architecture

This session only. Nothing lands on disk.

referenceschecklists.md

≈2.4k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Checklists — Design Systems Frontend Architecture

Use these checklists before finalizing design-system, UI critique, frontend implementation, or documentation work.

Discovery and context checklist

  • The product or feature goal is stated or a recommended default is declared.
  • The primary user task is clear.
  • The target audience and likely expertise level are identified.
  • The target platform/surface is known or assumed explicitly.
  • Existing system status is known: none, emerging, mature, fragmented, or migrating.
  • Existing constraints are captured: stack, CMS, framework, browser/device support, accessibility standard, brand guidelines, and release process.
  • The scope is framed by components, states, templates, content types, and behaviors rather than page count.
  • Known success criteria are listed: speed, consistency, accessibility, conversion, task completion, maintainability, or adoption.
  • Unknowns that materially change the recommendation are asked with a recommended default.
  • Routine best-practice questions are not asked.

Interface inventory checklist

  • Repeated UI is grouped by purpose, not only by visual similarity.
  • Repeated style treatments are grouped as perceptual patterns: color, type, spacing, shape, motion, imagery, tone.
  • Component candidates include anatomy, content model, actions, states, variants, and responsive behavior.
  • Visual duplicates with different purposes are separated.
  • Visually different elements with the same purpose are evaluated for consolidation.
  • One-off exceptions are flagged for removal, promotion to a variant, or documentation as an experiment.
  • Existing names are collected from design files, code, documentation, and team language.
  • Naming conflicts or ambiguous terms are listed for glossary cleanup.
  • High-frequency and high-risk patterns are prioritized first.

Design and UX quality checklist

  • The primary action is visually stronger than secondary and tertiary actions.
  • Labels describe user outcomes, not internal system terms.
  • Related elements are grouped and separated from unrelated elements.
  • Hierarchy works without relying on color alone.
  • Components communicate status, available actions, and consequences.
  • Empty states explain what happened and what the user can do next.
  • Loading states preserve layout stability where possible.
  • Error states identify the problem, location, and recovery path.
  • Validation is timed appropriately for the task and does not interrupt prematurely.
  • Content has been tested with long labels, missing data, dense data, and realistic examples.
  • The design does not solve a local issue by creating a global inconsistency.
  • The design includes states for hover, focus, active, disabled, selected, loading, success, warning, and error where applicable.

Component contract checklist

  • Component name is based on purpose or role, not appearance.
  • Purpose and user problem are stated.
  • Use cases and non-use cases are documented.
  • Anatomy identifies required and optional parts.
  • Variants are justified by role, emphasis, behavior, or context.
  • Props/API map to meaningful decisions rather than arbitrary styling escape hatches.
  • Content guidance covers length, truncation, empty values, icons, imagery, and localization risk.
  • Responsive behavior is specified at the component level.
  • Accessibility behavior is specified: semantics, labels, keyboard, focus, ARIA if needed.
  • The component has examples in isolation and inside at least one realistic composition.
  • The component has test guidance for important states and variants.
  • Ownership and contribution path are identified.

Token checklist

  • Token categories include color, typography, spacing, radius, border, elevation, motion, opacity, layer/z-index, size, and state where relevant.
  • Primitive tokens are separated from semantic tokens.
  • Semantic tokens are named by role and state, not raw appearance.
  • Component tokens exist only when semantic tokens are insufficient for controlled local mapping.
  • The token set avoids single-use tokens unless a one-off is explicitly documented as temporary.
  • Dark mode, theming, brand variants, and platform variants are represented only if needed.
  • Tokens include accessibility-sensitive states such as focus, disabled, error, warning, success, and selected.
  • Token names are stable enough to survive value changes.
  • Token documentation explains usage, not just values.
  • Deprecated tokens have replacements and migration notes.

Layout and responsive checklist

  • Layout is based on content needs and container behavior, not fixed page screenshots.
  • Components work at narrow widths, wide widths, high zoom, and large text.
  • Breakpoints are introduced only where readability, hierarchy, navigation, or task completion breaks.
  • Flexible layout primitives are used before fixed positioning or fixed dimensions.
  • Wrapping, stacking, truncation, and overflow behavior are specified.
  • Tables or data-dense views have horizontal overflow, column priority, or alternative layout guidance.
  • Sticky/fixed elements do not obscure content or focus targets.
  • The reading order remains logical when layout changes.
  • Images and media have aspect-ratio and fallback behavior.
  • Layout has been tested with realistic and edge-case content.

Accessibility checklist

  • Semantic HTML/native controls are used where available.
  • Interactive elements have accessible names.
  • Keyboard users can reach and operate every interactive control.
  • Focus order matches the visual/logical task order.
  • Focus indicators are visible and meet contrast expectations.
  • Text and meaningful UI indicators meet contrast requirements.
  • State is not communicated by color alone.
  • Error messages are associated with fields and announced where appropriate.
  • Dynamic content changes use live regions only when needed.
  • Touch/click targets are large enough or have extended hit areas.
  • Disabled states remain understandable and do not trap users.
  • Custom controls preserve native semantics, keyboard behavior, and form behavior.
  • Motion is brief, purposeful, and honors reduced-motion preferences.
  • Text remains readable under zoom, text resizing, and localization.

CSS and frontend implementation checklist

  • Markup is semantic and avoids unnecessary wrapper elements.
  • CSS selectors are low-specificity and component-scoped where appropriate.
  • Styling does not depend on deep page-location selectors.
  • Tokens are used for recurring values.
  • There is a clear CSS layering or naming convention.
  • !important is not used except in intentional, documented override layers.
  • Layout uses modern CSS primitives when supported by constraints.
  • JavaScript is used for behavior/state, not for simple presentational effects CSS can handle.
  • Dependencies are justified by repeated value, not one isolated effect.
  • Components include loading, error, empty, disabled, selected, and overflow states.
  • Build workflow includes formatting/linting and relevant tests.
  • Visual regression is recommended for high-value components or templates.
  • Package/version/distribution strategy is documented when components are shared.

Documentation checklist

  • Documentation states the pattern’s purpose.
  • Documentation includes “Use when” and “Do not use when.”
  • Anatomy, variants, states, content rules, accessibility, responsive behavior, and implementation notes are included.
  • Examples use realistic content.
  • Anti-examples show common misuse when helpful.
  • Related tokens and components are linked or named.
  • Design and code names match or are mapped clearly.
  • Ownership, contribution, review, and deprecation processes are stated.
  • Documentation is close enough to code/design source-of-truth to avoid drift.
  • The glossary captures ambiguous or high-frequency terms.

Governance checklist

  • There is an explicit owner or review group.
  • New patterns require evidence: repeated need, user problem, accessibility requirement, or implementation simplification.
  • Variants require purpose, not stylistic preference alone.
  • Experiments have a graduation, revision, or removal path.
  • Deprecations include replacement guidance.
  • Contribution process is clear for designers, developers, and product partners.
  • Adoption is measured or observable.
  • System health is reviewed periodically for drift, duplication, accessibility, and documentation staleness.

Critique final response checklist

  • The critique starts with the highest-impact issues, not minor polish.
  • Each issue includes why it matters and how to fix it.
  • Recommendations distinguish user-facing impact from implementation/system impact.
  • The response notes what is unknown and uses defaults where appropriate.
  • Accessibility concerns are integrated, not isolated at the end.
  • Suggested changes are concrete and testable.
  • The response avoids unsupported claims and avoids inventing product context.

Generation final response checklist

  • The response includes the system layer being defined: tokens, primitives, components, templates, pages, or governance.
  • Defaults are explicit and opinionated.
  • Context-dependent decisions are asked only when necessary.
  • The implementation guidance includes semantics, CSS/layout strategy, states, accessibility, and maintainability.
  • Documentation and governance are included when the work affects a reusable system.
  • Tradeoffs and override conditions are explained.
  • The next action is clear.

Source: SKILL.md on GitHub

No alerts16d3 checks · Risk SAFE
  • Gen Agent Trust Hub16d

    The skill is generally safe but carries a minor risk of indirect prompt injection and metadata poisoning as it processes external design assets and requirements without explicit sanitization or boundary markers.

  • Socket16d

    No alerts

  • Snyk16d

    Risk: LOW · No issues

Signed by skilld at 2841c07. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 2 months ago
metadata
{
  "author": "hueyexe"
}

README badge

README badge for hueyexe/frontend-agent-skills/design-systems-frontend-architecture