Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/f0d010c/stark/windows-designnpx skills add f0d010c/stark --skill windows-designgit clone --depth 1 https://github.com/f0d010c/starkWrote 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/f0d010c/stark/windows-design)<a href="https://agentmods.dev/skills/f0d010c/stark/windows-design"><img src="https://agentmods.dev/badge/skills/f0d010c/stark/windows-design.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 | $0.00142 | $0.02653 |
| Opus 5 | $0.00071 | $0.01326 |
| Sonnet 5 | $0.00028 | $0.00531 |
| Haiku 4.5 | $0.00014 | $0.00265 |
Grade A, and why
windows-design 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 4d 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 — 160 lines — stays where its author put it; the contents beside it link to each section on GitHub.
windows-design — pick the track first
Windows desktop has four distinct tracks. Each has different visual ceilings, different cost. Ask the user which one before any code. Do not default.
What This Skill Can Do
- Choose the right Windows track: strict WinUI, branded WinUI, Tauri, or Electron.
- Design desktop archetypes such as command centers, libraries, workbenches, monitoring cockpits, tray utilities, media apps, document apps, and settings/preferences.
- Preserve UX briefs, state coverage, command hierarchy, keyboard paths, inspector/detail panes, and recovery flows.
- Apply Fluent, Mica/Acrylic, Segoe UI Variable, Fluent icons, SettingsCard, NavigationView, InfoBar, and Windows density rules.
- Avoid generic sidebar/cards/table shells by using product-specific composition and anti-default rewrites.
Step 1 (MANDATORY) — Ask the user which track
Present these four options verbatim. Do not pick for them.
Which track do you want for this app?
1. System-like native (WinUI 3 strict) — feels like Settings/Calculator/Notepad. Best for: productivity utilities, internal tools, system panels. Visual ceiling: medium. Binary ~30MB. Users who care: Windows power users, IT admins. Examples: PowerToys, Snipping Tool, Sound Recorder.
2. Branded native (WinUI 3 + branded Fluent) — native chrome (Mica, Snap Layouts, ThemeResource auto dark/light/HC) but bespoke content surface (custom accent, custom display font, hero atmospheres, magazine layouts). Best for: consumer apps that want fit-in and identity. Visual ceiling: high but XAML-bound. Binary ~30MB. Examples: Apple Music for Windows, Files (community), Microsoft Photos rewrite, DevHome.
3. Tauri 2 + React/Tailwind/Motion — same React code as Electron but uses Windows' built-in WebView2 instead of bundled Chromium. Full visual freedom (mesh gradients, custom motion, real imagery, anything CSS can do). Binary ~5–10MB. RAM ~30–80MB. Mica still available via Tauri API. Examples: Cap, Spacedrive, ClashVerge, Pot.
4. Electron + React/Tailwind/Motion — bundles full Chromium. Same dev experience as Tauri but heavier. Binary ~150–250MB. RAM ~300–500MB. Largest ecosystem, every dev knows it. Examples: Spotify, Discord, VSCode, Slack, Notion, Figma desktop, Claude desktop, ChatGPT desktop.
Which? Or describe priorities (binary size, visual ambition, brand vs fit-in) and I'll pick.
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.
- 4d ago First seen · 160 lines · 142 tokens per session scan A c2fe01e49ec2
windows-design is a skill published in the GitHub repository f0d010c/stark (25 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 142 tokens to every session and 2,653 once invoked, about $0.0007 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 2026-08-30.
Other skills, from other repositories
assets
Generate, refine and integrate all landing media via Higgsfield MCP from DESIGN.md-derived briefs — hero images, textures, background loops, product shots. Use when craft invokes the asset phase, standalone to (re)generate one slot ("/award-craft:assets hero"), or to refine a slot from feedback ("/award-craft:assets…
frontend-design-writer
阶段 1-7 持续维护的 design.md 主写手——把调研（六维试吃）+ 双轮深度采访 + 视觉锚定三件套 + 动效实现 prompt 集全部压缩为 Agent 可执行的设计契约。kun 版 15 段三轮写作顺序（原 12 段 + 新增 §12 Moodboard Prompt / §13 Page Mockup Prompt / §14 Motion Prompts 三段视觉锚定引用）。同时维护 DESIGN-LANGUAGE.md（低频）+ design.md（高频）+ MOTION-SPEC.md（中频）三份。不用于调研本身（那是 frontend-design-research）、不用于从零搭 knowledge/…
frontend-visual-reference
阶段 4 视觉锚定三件套的主 skill——在写第一行代码前，用 Moodboard（A） + 完整网页示意图（B） + 动效帧序列（C） 三种图像级锚点把视觉方向钉死。探索前移在 Moodboard 阶段（A）以 2-3 张让用户选的方式执行，B/C 是单线生成。资产生成走双路径：主路径 baoyu-imagine skill 自动出图，兜底路径打开 prompt 文件让用户复制到网页端。产出写入 knowledge/design-docs/visual-anchors/ 目录，并在 design.md §12/§13/§14 回填引用。触发：design.md 初稿完成、后轮采访定稿后、用户说"出几张方向图让我选""做…
frontend-design-research
阶段 2 试吃——对 4-5 个该领域设计优秀的真实网站跑六维机械拆解（形 / 色 / 字 / 构 / 质 / 动），综合成本项目 design.md 初稿。六维是上位骨架，DOM 量化工作流 + 动效采访 5 类是落地肌肉。每次一站，保护上下文。不做前轮/后轮深度采访（那是 frontend-interview-dualround）、不做视觉锚定三件套（那是 frontend-visual-reference）、不写最终交付的 design.md（那是 frontend-design-writer）、不做组件库 / 动效资产库选型（那部分由 frontend-design-writer 的 §11 Resources 段 +…
frontend-design-review
代码已经跑起来后（阶段 7 迭代微调）的设计评审横切——用户看到具体页面涌现审美反馈时使用：三段式追问（定位→对标→量化）把"感觉不对"翻译成可执行改动项，"选择题化"让用户做选择题而不是填空题，参数外显 Tweaks 让不确定的参数暴露成可调旋钮，所有修正反写回 design.md。覆盖 design-review（视觉对标）、design-content（文案与信息层级）、design-harden（边缘状态加固）三种子模式。仅用于代码已生成后——代码前的"感觉"引导走 frontend-interview-dualround 的后轮采访；纯调研走 frontend-design-research；写 design.md 走…
frontend-i18n-essentials
前端国际化（i18n）的综合规划与执行指南——做国际性网站、多语言架构、补多语言、翻译质量验证、字体跨语言统一、路由 / SEO / 字符串提取 / 动态切换等 i18n 相关任务时用。本 skill 按"运行时 / 翻译内容 / 视觉"三层正交模型路由：先定位用户的 i18n 痛点在哪一层，再走对应决策流；翻译内容层默认走三段流水线（机翻 → AI 本地人审 → 人工兜底）+ subagent 并行化（locale × persona × domain）。触发：用户说"做国际性网站""加多语言""补英文版""做 i18n""多语言网站的 SEO""路由用子路径还是子域名""本地化翻译""文化陷阱""AI…