Token导航 LogoToken导航TokenDH.com
研究检索操作浏览器github未标认证来源可访问许可证需确认审计通过

design-flow设计流程

Agent Skill

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。它适合让 Agent 根据产品场景整理页面结构、生成 UI 方案、检查视觉一致性或改进组件层级。使用时需要结合现有品牌、设计系统和用户任务,不应只堆装饰元素;涉及真实页面改动时,应通过截图或浏览器预览检查文本溢出、对齐和响应式表现。

总安装

23,338

周安装

935

GitHub Stars

88

下载量

7,555
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:design-flow(设计流程)
来源仓库:https://github.com/julianoczkowski/designer-skills
仓库路径:skills/design-flow
安装命令:
npx skills add https://github.com/julianoczkowski/designer-skills --skill design-flow
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/julianoczkowski/designer-skills --skill design-flow

简介

串联完整设计师工作流,依次执行澄清思路、撰写简报、架构信息等阶段。

  • 适合新手设计师系统学习全流程方法,资深者也可用于流程管控。
  • 每个环节必须达成明确共识方可进入下一阶段,避免无效劳动。
  • 提供设计令牌、任务分解等关键产出物,形成可追溯的知识资产。
  • design-flow 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

This skill orchestrates the full designer workflow by running each skill in sequence. You are a guide walking the designer through each phase. Do not rush. Each phase must be completed and confirmed before moving to the next.

Example prompts

  • "Run the full design flow"
  • "Walk me through the complete process for a new project"
  • "Start from scratch and take me through everything"
  • "Design flow for a dashboard app"

The Sequence

1. Grill Me          → Clarify thinking
2. Design Brief      → Document intent
3. Info Architecture  → Define structure
4. Design Tokens     → Establish visual system
5. Brief to Tasks    → Plan the build
6. Frontend Design   → Build it
—
7. Design Review     → Run separately when ready

Rules

  1. At the start, tell the designer what the full sequence looks like (phases 1-6, with review available separately) and ask if they want to skip any phases. Common skip patterns:

- Already have a clear idea → skip grill-me - Single component, not a full page → skip information-architecture - Existing project with tokens → skip design-tokens

  1. Before each phase, announce which phase you are entering and what it will produce. Example: "Phase 2: Design Brief. I'll interview you about the project and produce a DESIGN_BRIEF.md file. Ready?"
  2. During each phase, read the corresponding SKILL.md file and follow its full instructions. Do not summarize or abbreviate the skill. Run it properly.
  3. After each phase, summarize what was produced (the file name, the key decisions, any open questions) and ask: "Ready to move to the next phase?" Wait for confirmation.
  4. Between phases, check if the output from the previous phase changes anything about the next phase. For example, if the brief names a philosophy, mention that the tokens phase will use it.
  5. The designer can stop at any point. If they say "that's enough for now," summarize where they are in the sequence and what the next phase would be when they return.

Phase Details

Phase 1: Grill Me

Read the grill-me skill (grill-me/SKILL.md) and follow its instructions. Produces: Shared understanding of the project. No file output. Transition: "We've resolved the key decisions. Ready to capture this as a design brief?"

Phase 2: Design Brief

Read the design-brief skill (design-brief/SKILL.md) and follow its instructions. Produces: .design/<feature-slug>/DESIGN_BRIEF.md. Transition: "The brief is saved. Next is information architecture, where we'll define the page structure and navigation. Skip this if you're building a single component. Continue?"

Phase 3: Information Architecture

Read the information-architecture skill (information-architecture/SKILL.md) and follow its instructions. Produces: .design/<feature-slug>/INFORMATION_ARCHITECTURE.md. Transition: "IA is defined. Next we'll generate design tokens (colors, spacing, typography) based on the philosophy from the brief. Continue?"

Phase 4: Design Tokens

Read the design-tokens skill (design-tokens/SKILL.md) and follow its instructions. Produces: Token file (CSS variables, Tailwind config, or theme file depending on stack). Transition: "Tokens are set. Next I'll break the brief into a task list so we can build in order. Continue?"

Phase 5: Brief to Tasks

Read the brief-to-tasks skill (brief-to-tasks/SKILL.md) and follow its instructions. Produces: .design/<feature-slug>/TASKS.md. Transition: "Tasks are ready. Now we build. I'll start with the first task on the list. Continue?"

Phase 6: Frontend Design

Read the frontend-design skill (frontend-design/SKILL.md) and follow its instructions. Work through the tasks from TASKS.md in order. After completing each task, check it off and confirm with the designer before moving to the next task. Produces: Built components and pages. Transition: "The flow is complete. Your brief, IA, tokens, and tasks are all saved in the project. When you're ready for a design review, run /design-review and I'll critique the build against the brief."

The flow ends here. Phase 7 is not automatic.

Phase 7: Design Review (on request only)

This phase does NOT run automatically. It only runs if:

  • The designer explicitly asks for a review during the flow
  • The designer runs /design-review separately after building

The review requires built code to examine. If no components or pages have been built yet, do not run this phase. Instead, remind the designer: "Run /design-review once you have something built. It will check the output against the brief."

When triggered, read the design-review skill (design-review/SKILL.md) and follow its instructions. The review will capture screenshots of the running application using Playwright MCP (preferred), the Cursor IDE Browser (fallback), or by asking the user to provide them manually if no browser tool is available.

Produces: .design/<feature-slug>/DESIGN_REVIEW.md + screenshots saved in .design/<feature-slug>/screenshots/. Transition: "Review is done. Screenshots are saved in .design/<feature-slug>/screenshots/. If there are must-fix items, I can address them now."

Project Files Structure

All design flow artifacts are saved under .design/<feature-slug>/ where <feature-slug> is a short, lowercase, hyphenated name derived from the feature being designed. This ensures multiple features can be designed independently without overwriting each other.

.design/
└── <feature-slug>/
    ├── DESIGN_BRIEF.md              ← Phase 2: Project intent, goals, aesthetic direction
    ├── INFORMATION_ARCHITECTURE.md  ← Phase 3: Navigation, page structure, user flows
    ├── DESIGN_TOKENS.*              ← Phase 4: Colors, spacing, typography, shadows (CSS/Tailwind/theme)
    ├── TASKS.md                     ← Phase 5: Ordered build checklist from the brief
    ├── DESIGN_REVIEW.md             ← Phase 7: Prioritized critique against the brief
    └── screenshots/                 ← Phase 7: Visual evidence from the running app
        ├── review-[page]-desktop-1280.png
        ├── review-[page]-tablet-768.png
        ├── review-[page]-mobile-375.png
        ├── review-[page]-dark-mode-*.png
        └── review-[component]-[state].png

The screenshots/ subfolder is created during the design review phase. All visual evidence of the review (responsive breakpoints, interactive states, dark mode) is saved here with descriptive filenames so findings in DESIGN_REVIEW.md are traceable.

If the Designer Returns Mid-Flow

Check the .design/ folder for existing feature subfolders. If files from earlier phases exist (DESIGN_BRIEF.md, INFORMATION_ARCHITECTURE.md, TASKS.md) inside a feature folder, read them to understand where the designer left off. Ask which feature to resume if multiple folders exist. Resume from the next incomplete phase.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.01%
按下载量换算2,721

Claude

28.96%
按下载量换算2,188

Cursor

19.01%
按下载量换算1,436

Gemini CLI

9.28%
按下载量换算701

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills