Skip to main content
reopt Handbook
reopt Handbook
Design Systems for the AI Era

Foundation

Token ArchitectureComponent SpecBuilt-in Accessibility

AI Workflow

Prompt InterfaceDESIGN.md InterfaceAgentic Design Quality ControlWorkflow StrategiesAI-Based Design System EvolutionContext Injection

User Experience

Consistency PatternsInteraction DesignForms and Data Input

Next Interfaces

Agent UI ProtocolsGenerative UISpatial and Multimodal Interfaces

Practice

Execution PlaybookMigration StrategyGovernance and CollaborationCase Studies

Operations

Verification ChecklistUpdate Log
Handbook›Design Systems for the AI Era›Verification Checklist
한국어English

Verification Checklist

Quality checks for bilingual structure, sources, and AI-ready design-system guidance.

This page defines how the handbook itself is verified. Design-system documentation is not a static artifact. It needs recurring correction as AI coding tools, design tools, and token standards change.

Automated Checks

Run these commands after handbook changes.

pnpm --filter handbook run check:design-systems-ai
pnpm --filter handbook run typecheck

check:design-systems-ai verifies:

AreaRule
Page structureKorean and English meta.json page order must match.
File coverageEvery registered .mdx page must exist in both locales.
FrontmatterEvery page must include title and description.
Link localizationEnglish pages must use /en/books/design-systems-ai/ links, not /books/design-systems-ai/.
Verification coverageThis page must reference every primary chapter slug.

Chapter Coverage

slugReview focus
token-architectureConfirm DTCG, Style Dictionary, and Tailwind CSS token guidance against current sources.
component-specConfirm props, slots, states, variants, and accessibility contracts are machine-readable.
accessibilityConfirm ARIA, keyboard, focus, and accessible-name rules are implementable.
prompt-interfaceConfirm prompt examples include component choice, forbidden patterns, and verification commands.
design-md-interfaceConfirm DESIGN.md communicates brand intent and visual rules clearly.
agentic-design-controlConfirm screenshot review, Playwright checks, and agent feedback loops are explicit.
workflow-strategiesConfirm design-to-code, code-to-design, and bidirectional workflows are not overstated.
ds-evolutionConfirm AI-assisted refactoring, audits, and documentation expansion are operational.
context-injectionConfirm AGENTS.md, CLAUDE.md, DESIGN.md, and MCP responsibilities are separated.
consistencyConfirm grid, spacing, density, and responsive behavior are written as verifiable rules.
interactionConfirm feedback, motion tokens, loading, error, and empty states are covered.
formsConfirm validation, error copy, field composition, and accessibility are treated together.
agent-ui-protocolsConfirm A2UI, AG-UI, MCP Apps, and similar claims match public documentation.
generative-uiConfirm runtime UI generation is constrained by schemas, registries, and guardrails.
spatial-multimodalConfirm spatial, voice, and multimodal rules connect back to design-system tokens.
playbookConfirm the 0 -> 1 sequence is small enough for teams to execute.
migrationConfirm migration is incremental rather than a big-bang rewrite.
governanceConfirm ownership, approval, review, and quality gates connect to docs and code.
case-studiesConfirm examples are framed as reusable patterns and limits, not tool promotion.

Source Refresh Rules

Check version and date before making status claims.

TopicSource rule
Design TokensCheck the latest Design Tokens Format Module report and the DTCG site.
Style DictionaryCheck official Style Dictionary docs before claiming full 2025.10 DTCG support.
Tailwind CSSCheck official Tailwind docs for v4 @theme and theme variables.
Figma MCPCheck Figma developer docs and help center for remote/desktop MCP status and beta wording.
ReactCheck the official React release blog for React 19.x feature status.
MCP AppsCheck the Model Context Protocol blog before using proposal, extension, or production-ready wording.

Manual Review

Automation protects structure, not judgment. For meaningful changes:

  1. Confirm the Korean and English pages make the same technical claims.
  2. Confirm English pages have enough context for global readers, not only literal translation.
  3. Check that "official", "GA", "stable", "beta", and "production-ready" match the cited source.
  4. Pair every AI workflow recommendation with validation loops, failure modes, and explicit constraints.

Related docs

Update Log

Change history and next improvement candidates for the AI-era design systems handbook.

Verification Report

Developer Unlearning · Verification notes for the English Developer Unlearning handbook.

Cmd. /permissions

Codex Command Master · Change Codex approval and permission policy during a session.

Renewal and Risk

B2B SaaS Sales and Customer Success · Detect renewal risk early and operate a structured recovery process.

Update Log

Developer Unlearning · English Developer Unlearning handbook change log.

Case Studies

Practical patterns from shadcn, Radix, Stitch, Material, Fluent, and AI-first systems.

Update Log

Change history and next improvement candidates for the AI-era design systems handbook.

On this page

Automated ChecksChapter CoverageSource Refresh RulesManual Review