본문으로 건너뛰기
cccskills
무료GitHub에 공개

shape

Plan the UX and UI for a feature before writing code. Run a structured discovery interview, then produce a design brief (.design-context.md) that guides implementation. Establishes design direction, constraints, and strategy before any code is written. Triggers on plan the design, design brief, shape the UX, before we build, design discovery.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(3)

  • SKILL.md7.5 KB
  • README.md278 B
  • references/design-media-catalog.md2.9 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

Shape the UX and UI for a feature before any code is written. This skill produces a design brief — a structured artifact that guides implementation through discovery, not guesswork.

Scope: Design planning only. This skill produces design briefs, not code.

Preparation

Read ~/.claude/skills/minoan-frontend-design/SKILL.md for aesthetic principles and anti-pattern guidance. These inform the design direction but should not constrain the discovery interview.

Check for existing design context: .design-context.md in the project root, or a ## Design Context section in CLAUDE.md. If either exists, use it as a starting point — don't re-ask questions it already answers.

Phase 1: Discovery Interview

Do NOT make any design decisions during this phase. Understand the feature deeply enough to make excellent decisions later.

Interview cadence

Ask 2-3 questions per round, then wait for answers. One round is the default. Add a second only if the first answers leave material gaps — don't run a second round just to feel thorough.

Assert-then-confirm, not menu-with-escape. When the user's prompt and existing context make one option obvious, name it and ask the user to confirm or override. "This reads as Restrained, confirm?" beats a four-option menu when the answer is already clear.

Ask these questions conversationally — adapt based on answers, don't dump them all at once:

Purpose & Context

  • What is this feature for? What problem does it solve?
  • Who specifically will use it? (Role, context, frequency — not just "users")
  • What does success look like? How will you know this feature is working?
  • What's the user's state of mind when they reach this? (Rushed? Exploring? Anxious?)

Content & Data

  • What content or data does this feature display or collect?
  • What are the realistic ranges? (Minimum, typical, maximum — e.g., 0 items, 5 items, 500 items)
  • What are the edge cases? (Empty state, error state, first-time use, power user)

Design Direction

Force a visual decision on three fronts. Skip anything existing context already answers; ask only what's missing.

  • Color strategy for this surface. Pick one: Restrained / Committed / Full palette / Drenched. Can override the project default if the surface earns it (e.g. a drenched hero inside an otherwise Restrained product).
  • Theme via scene sentence. Write one sentence of physical context for this surface: who uses it, where, under what ambient light, in what mood. The sentence forces dark vs light. If it doesn't, add detail until it does.
  • Two or three named anchor references. Specific products, brands, objects. Not adjectives like "modern" or "clean."
  • What's the single most important thing a user should do or understand here?
  • Are there existing patterns in the product this should be consistent with?

Scope

Always ask. Sketch quality and shipped quality are different outputs — don't guess between them.

  • Fidelity. Sketch / mid-fi / high-fi / production-ready?
  • Breadth. One screen / a flow / a whole surface?
  • Interactivity. Static visual / interactive prototype / shipped-quality component?
  • Time intent. Quick exploration, or polish until it ships?

Constraints

  • Technical constraints? (Framework, performance budget, browser support)
  • Content constraints? (Localization, dynamic text length, user-generated content)
  • Mobile/responsive requirements?
  • Accessibility requirements beyond WCAG AA?

Anti-Goals

  • What should this NOT be? What would be a wrong direction?
  • What's the biggest risk of getting this wrong?

Phase 1.5: Visual Direction Probe (Capability-Gated)

After the interview, generate 2-4 visual direction probes before writing the final brief when all of these are true:

  • The work is net-new or directionally ambiguous enough that visual exploration will clarify the brief.
  • The requested fidelity is mid-fi, high-fi, or production-ready.
  • The current harness gives you native image generation (e.g. image_gen, an MCP tool, or similar).

When those conditions are met, this step is mandatory. If image generation isn't natively available, state in one line that the step is skipped and proceed.

The probes should differ in primary visual direction (hierarchy, topology, density, typographic voice, color strategy), not just palette tweaks. Treat them as direction tests — ask the user which feels closest, what's off, and what should carry forward. If probes reveal a mismatch, revise the brief inputs before finalizing.

Phase 2: Design Brief

After the interview (and any probes), synthesize into a structured design brief. Present it and stop — the user must confirm before considering this skill complete.

Choose the brief shape based on clarity:

  • Compact form (3-5 bullets) when discovery was crisp and existing context already pins scope, content, and direction. End with one or two specific questions or "confirm or override?"
  • Full structured form when the task is genuinely ambiguous, multi-screen, or the user asked for shape as a standalone step.

Don't pad a clear brief into a long one to look thorough.

1. Feature Summary (2-3 sentences) What this is, who it's for, what it needs to accomplish.

2. Primary User Action The single most important thing a user should do or understand here.

3. Design Direction Color strategy (Restrained / Committed / Full palette / Drenched) + theme scene sentence + 2-3 named anchor references. Reference the project's design context if available. Set DESIGN_VARIANCE, MOTION_INTENSITY, and VISUAL_DENSITY (1-10 each) based on discovery answers. Consult ~/.claude/skills/minoan-frontend-design/references/design-dials.md. If Visual Direction Probes were run, name which direction won and what changed.

4. Layout Strategy High-level spatial approach: what gets emphasis, what's secondary, how information flows. Describe the visual hierarchy and rhythm, not specific CSS.

5. Key States List every state the feature needs: default, empty, loading, error, success, edge cases. For each, note what the user needs to see and feel.

6. Interaction Model How users interact with this feature. What happens on click, hover, scroll? What feedback do they get? What's the flow from entry to completion?

7. Content Requirements What copy, labels, empty state messages, error messages, and microcopy are needed. Note any dynamic content and its realistic ranges.

8. Open Questions Anything unresolved that the implementer should resolve during build.

9. Implementation Skills Consult references/design-media-catalog.md. Based on discovery answers, recommend which skills the implementer should invoke beyond /minoan-frontend-design. Only list skills that match actual needs discovered — don't enumerate the full catalog.


Get explicit confirmation of the brief before finishing. If the user disagrees with any part, revisit the relevant discovery questions.

Once confirmed, write the brief to .design-context.md in the project root (or update the existing file). The user can then hand it to /minoan-frontend-design or use it to guide any other implementation approach.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

Search academic papers, build literature reviews, and synthesize research findings — combines Exa search (publication category, domain and date filters) with arxiv-mcp-server for paper discovery, download, and deep analysis. Triggers on academic paper, literature review, research synthesis, arxiv, find papers, scholarly search.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

This skill should be used when creating, auditing, or maintaining AGENTS.md files and Codex CLI configuration for any project—including initializing AGENTS.md for cross-agent compatibility (Codex, Cursor, Copilot, Devin, Jules, Amp, Gemini CLI), generating config.toml or .rules files, scaffolding .agents/skills/, converting CLAUDE.md to AGENTS.md, or auditing existing agent configs for bloat and staleness. Complementary to codex-orchestrator (which executes subagents; this skill creates the config files they consume).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

Academic research skill for Biblical Hebrew, Semitic linguistics, cuneiform studies, and comparative Ancient Near Eastern research. Provides Sefaria API for Hebrew Bible, CDLI/ORACC for cuneiform databases, and web discovery via Omnisearch, Exa, Firecrawl, and Obscura for finding scholarly sources across JSTOR, Perseus, Persée, Google Scholar, and academia.edu. Triggers on Hebrew quotes, cuneiform, Sefaria, ANE research, Minoan, search for scholarship, find papers, literature review, scholarly search, academic search, Genesis/Tehom, Ugaritic, Talmudic sources, extract from PDF, OCR academic.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

Build comprehensive ARCHITECTURE.md files following matklad's canonical guidelines — bird's-eye views, ASCII/Mermaid diagrams, codemaps, invariants, and layer boundaries. Triggers on document the architecture, create ARCHITECTURE.md, map this codebase, architectural overview.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

Generate physically-based atmospheric scattering shaders — sky domes, planetary atmospheres, LUT-optimized pipelines, depth-aware post-processing. Four modes — sky-dome, atmosphere-post, planet, lut. Triggers on atmospheric scattering, sky shader, sunset rendering, planet atmosphere, Rayleigh scattering, Mie scattering, volumetric sky, sky dome, atmosphere post-processing, aerial perspective, transmittance LUT, sky rendering, realistic sky, planetary rendering.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

tdimino/claude-code-minoan412026년 10월 20일 업데이트

tdimino의 스킬 모두 보기

이 스킬의 문제 신고