Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add kkunkunya/ai-frontend-design-kit/plugin install ai-frontend-design-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kkunkunya/ai-frontend-design-kit/frontend-visual-reference)<a href="https://agentmods.dev/skills/kkunkunya/ai-frontend-design-kit/frontend-visual-reference"><img src="https://agentmods.dev/badge/skills/kkunkunya/ai-frontend-design-kit/frontend-visual-reference.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00251 | $0.05714 |
| Opus 5 | $0.00125 | $0.02857 |
| Sonnet 5 | $0.00050 | $0.01143 |
| Haiku 4.5 | $0.00025 | $0.00571 |
Grade A, and why
frontend-visual-reference scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 355 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Packaged Knowledge Snapshot
This packaged copy includes local note snapshots for portability. When the original Obsidian absolute path is unavailable, use the packaged snapshot instead:
/Users/kunkun/note/01-AI工程/AI设计与前端/01-认知框架/AI前端设计认知框架.md->references/packaged-knowledge/AI前端设计认知框架.md/Users/kunkun/note/01-AI工程/AI设计与前端/01-认知框架/六维：形色字构质动是网站试吃的机械骨架.md->references/packaged-knowledge/六维：形色字构质动是网站试吃的机械骨架.md/Users/kunkun/note/01-AI工程/AI设计与前端/02-执行方法/动效文档两层：MOTION-SPEC vs 实现 prompt 集.md->references/packaged-knowledge/动效文档两层：MOTION-SPEC vs 实现 prompt 集.md/Users/kunkun/note/01-AI工程/AI设计与前端/02-执行方法/反 slop 硬禁令：AI 前端的十条视觉红线.md->references/packaged-knowledge/反 slop 硬禁令：AI 前端的十条视觉红线.md/Users/kunkun/note/01-AI工程/AI设计与前端/02-执行方法/审美是隐性认知，追问是唯一的解压工具.md->references/packaged-knowledge/审美是隐性认知，追问是唯一的解压工具.md/Users/kunkun/note/01-AI工程/AI设计与前端/02-执行方法/视觉锚定三件套：代价结构与资产生成双路径.md->references/packaged-knowledge/视觉锚定三件套：代价结构与资产生成双路径.md
Frontend Visual Reference
这个 skill 在做什么
阶段 4 视觉锚定三件套的执行器。在写第一行代码之前，用三种图像级锚点把视觉方向钉死：
| 锚点 | 图像形态 | 凝聚维度 | 探索属性 |
|---|---|---|---|
| A. Moodboard | 综合风格板（色板 + 材质样本 + 符号族 + 字体 + UI 片段） | 六维静态 5 维（形/色/字/构/质） | ★ 前移探索：2-3 张让用户选 |
| B. 完整网页示意图 | 主页 → 子页的完整页面示意图 | 真实页面布局 + 信息层级 + 叙事推进 | 单线：主页 → OK → 子页 |
| C. 动效帧序列 | 从左到右的帧变化示例 | 六维动维 | 单线：每个核心动效 1 套 |
核心赌局：AI 生图便宜 + 视觉传达直接，所以把"探索"前移到图像层，代码层单线执行。
在阶段化流程里的位置
阶段3 后轮深度采访 → design.md 定稿
↓
阶段4 视觉锚定三件套 ← 本 skill
├─ A. Moodboard Prompt（2-3 张，前移探索）
├─ B. 完整网页示意图 Prompt（主页 → 子页，单线）
└─ C. 动效帧序列 Prompt（每个核心动效 1 套，单线）
↓
阶段5 资产生成（baoyu-imagine 主 + 手动兜底）
↓
阶段6 代码实现（读 design.md + 三件套图像 + 动效微观 prompt）
关键重定位（和老版的差异）：老版 visual-reference 是"Layer 4 方向选择门"——派生 2-4 个视觉方向候选出图让用户选。2026-04-21 重构后明确：
- 老版"派生 2-4 视觉方向"吸收进本 skill 的 A. Moodboard 阶段——2-3 张 Moodboard 做探索分支
- B. 完整网页示意图、C. 动效帧序列 是新增职责
- "探索" 只在 A 阶段做，B/C 是单线执行（A 选定后方向已钉死）
- 产出路径从
visual-directions/改为visual-anchors/（三件套统一收纳）
What ships with it
7 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- references/packaged-knowledge/AI前端设计认知框架.md 25 KB
- references/packaged-knowledge/manifest.json 1.3 KB
- references/packaged-knowledge/六维：形色字构质动是网站试吃的机械骨架.md 8.5 KB
- references/packaged-knowledge/动效文档两层：MOTION-SPEC vs 实现 prompt 集.md 8.5 KB
- references/packaged-knowledge/反 slop 硬禁令：AI 前端的十条视觉红线.md 8.4 KB
- references/packaged-knowledge/审美是隐性认知，追问是唯一的解压工具.md 11 KB
- references/packaged-knowledge/视觉锚定三件套：代价结构与资产生成双路径.md 7.6 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 355 lines · 251 tokens per session scan A 51b80ba03766
frontend-visual-reference is a skill published in the GitHub repository kkunkunya/ai-frontend-design-kit (7 stars, last pushed 3mo ago), licensed MIT. It adds 251 tokens to every session and 5,714 once invoked, about $0.0013 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
stark
Use for UI/UX design, product flows, frontend or native interface design, design audits, design translation, asset planning, motion direction, design tokens, or distinctive non-generic apps and websites. Covers web and React, Windows and WinUI/Mica, Apple and SwiftUI, Android and Jetpack Compose/Material…
web-design
Use for web interfaces, landing pages, product sites, dashboards, React, Next.js, Astro, Tailwind, frontend audits, responsive design, browser interaction, web motion, or requests for a distinctive non-generic website. Select the web surface and implementation track, load a small reference bundle, preserve usability…
design-with-claude
Use when design work needs a product designer's eye: auditing a codebase for design-system gaps, fixing WCAG contrast and unlabeled inputs, choosing type scales or spacing steps, reviewing UI that looks generic or AI-generated, or designing forms, tables, dashboards, navigation, checkout, onboarding, dark mode, and…
web-prototype
General-purpose desktop web prototype. Single self-contained HTML file built by copying the seed assets/template.html and pasting section layouts from references/layouts.md. Default for any landing / marketing / docs / SaaS page when no more specific skill matches.
web-prototype-taste-brutalist
Swiss industrial-print web prototype. Newsprint canvas, monolithic black grotesque, viewport-bleeding numerals, hairline grid dividers, hazard-red accent, ASCII syntax decoration. Distilled from Leonxlnx/taste-skill brutalist-skill (Swiss Industrial Print mode).
web-prototype-taste-soft
Apple-tier soft web prototype. Silver/cream canvas, double-bezel cards, button-in-button CTAs, generous squircle radii, spring motion, ambient mesh. Distilled from Leonxlnx/taste-skill soft-skill + sections 4–8 of taste-skill.