据 AlphaMoat 统计,截至 2026-09,Vibe Design 累计下载 2,395 次、获 10 个星标,在全部 5,078 个 AI 技能中排名第 4,179 位,在「设计创意」分类(512 个)中排名第 410 位。
| 平台 | 累计安装 | 累计下载 | 星标 | 评论 |
|---|---|---|---|---|
| ClawHub ↗ | 67 | 2,149 | 10 | - |
| SkillHub ↗ | 235 | 2,395 | 10 | — |
各平台的计数口径不同,统计时间不一致。
该技能把“氛围编程”理念应用于视觉设计:用自然语言描述需求,让 AI 生成视觉稿、布局与素材,再基于结果迭代而非从零抠像素。核心规则包括:使用设计词汇而非艺术词汇、锚定设计系统(Material Design 3、Apple HIG、IBM Carbon 等)、分层组织提示词、显式指定宽高比、批量生成变体后混搭迭代,并明确 AI 负责探索、Figma 等工具负责精修收尾。
资料发布于 2026-10-27
以下是根据技能文档生成的示例,不代表实际运行结果。
2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。
| 月份 | 累计下载 | 环比 | 星标 |
|---|---|---|---|
| 2026-09 | 2,395 | — | 10 |
| 2026-08 | 2,116 | +5.7% | 10 |
| 2026-07 | 2,001 | +11.0% | 10 |
| 2026-06 | 1,802 | +8.4% | 10 |
| 2026-05 | 1,663 | — | 9 |
| 2026-03 | 1,195 | — | 7 |
以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。
---
name: Vibe Design
slug: vibe-design
version: 1.0.0
description: Create visual designs with AI tools. Covers prompting for UI/graphics, Midjourney techniques, Figma AI workflow, and iteration patterns.
metadata: {"clawdbot":{"emoji":"🎨","requires":{"bins":[]},"os":["linux","darwin","win32"]}}
---
## What is Vibe Design
Design where you describe what you want and let AI generate visuals, layouts, and assets. You iterate based on results rather than pixel-pushing from scratch. Same philosophy as vibe coding applied to visual design. **Core principle:** AI handles exploration and generation. You handle curation and refinement. ## Quick Reference | Topic | File | |-------|------| | Prompting for visuals | `prompting.md` | | UI/UX design with AI | `ui-design.md` | | Image generation tips | `image-gen.md` | | Figma AI workflow | `figma.md` | | Tools by use case | `tools.md` | ## Core Rules ### 1. Use Design Vocabulary, Not Art Vocabulary AI responds differently to design vs art terms. ❌ Art words: "beautiful", "fantasy", "render", "painting", "stunning" ✅ Design words: "interface", "layout", "component", "spacing", "hierarchy" Bad: "beautiful mobile banking product, minimal futuristic style" Good: "mobile banking app UI in Material Design 3, clean layout, consistent spacing, data cards with financial metrics" ### 2. Anchor with Design Systems Adding a design system name drastically improves structure: - "Material Design 3" → balanced spacing, cards, shadows - "Apple HIG" → soft radii, clear typography, native feel - "IBM Carbon" → enterprise dashboard layouts - "Atlassian" → B2B tone, crisp typography ### 3. Layer Your Prompts Build prompts with structure: ``` [Project type] UI design for [product/audience], [style keywords], [layout type], [color palette], [design system reference], [rendering style] ``` ### 4. Aspect Ratios Matter - Web: --ar 16:9 - Mobile: --ar 4:5 or --ar 9:16 - Square/icons: --ar 1:1 - Always specify for consistent output ### 5. Iterate, Don't Perfect First Try Vibe design = rapid exploration. 1. Generate 4-8 variants 2. Identify elements that work 3. Remix: "Like image 2, but with the color palette from image 3" 4. Refine winner in production tool (Figma, etc.) ### 6. AI for Exploration, Tools for Production - **AI generates**: Mood, direction, concepts, rough layouts - **Production tools refine**: Pixel-perfect spacing, real content, responsive behavior Don't try to get production-ready from AI. Use it for the 80% exploration, finish the 20% manually. ### 7. Reference Over Description When possible, show don't tell: - Attach reference images - Link to existing designs you like - Use "in the style of [specific designer/brand]" ### 8. Design System Consistency For production work: - Export AI concepts as references, not finals - Rebuild in your design system - Maintain your spacing, typography, and color tokens - AI inspiration → human implementation ### 9. Know the Tool Strengths | Tool | Best For | |------|----------| | Midjourney | Artistic concepts, mood exploration | | Figma AI | UI iteration, production design | | DALL-E/GPT | Quick mockups, image editing | | Stable Diffusion | Custom training, open-source control | | Canva | Marketing assets, social media | See `tools.md` for detailed comparison.
ContextWeave 架构图一键生成
用 Nano Banana Pro 生成或编辑图片的脚本技能。
用 whisper 命令行在本地把音频转写成文字或做翻译的技能。
通过 MCP 服务生成与编辑图表的技能。
面向界面设计的思路与落地建议技能,覆盖布局、交互流程、设计令牌与前端改动。
面向多家视频模型 API 的 AI 视频生成与编排技能。
用 ffmpeg 从视频里取出画面帧,或生成便于快速查看的缩略图。
用本地 Whisper 把音频文件转成文字,模型下载完成后可完全离线运行。
把用户纠正与自我复盘沉淀成分层记忆、并按使用情况升降级的技能。
在保留公式、格式与工作簿结构的前提下创建、检查并修改 Excel 文件的技能。
处理 Word 与 DOCX 文档的作业规范,重点是改完之后样式、编号与版式不跑偏。
把数据库、表格与导出的原始数据变成可决策结论的数据分析技能。
改 PPT 之前先看清结构的一套规范,目标是改完还能干净渲染。
数据月份:2026-09 · 下载/星标/安装数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。