Art & Design

Evaluate and plan website animation and interaction, then offer 2 to 3 comparable versions with a clickable mobile prototype before any production code changes. Use when the user asks for website animation, motion, transitions, page transitions, smooth or app-like feel, or complains that a site feels static, stiff, dated, like a template, or like it jumps between pages, even if they never say animation.

Create Spine skeletal animations from images with agent-authored part and motion plans, export JSON/atlas/PNG for game engines, and open a local animation demo page. Use for character animation, image rigging, blinking, or UI motion in Spine.

Monta landing pages com efeito que viraliza (personagem que segue o mouse, lanterna que revela o que tem por dentro, fundo em loop com parallax) gerando as imagens e o vídeo com o MCP do Higgsfield, escrevendo um prompt pixel perfect em oito partes e conferindo o resultado por print. Use quando pedirem um site, uma landing page ou um hero "com efeito", "cinematográfico", "pixel perfect", ou um desses três efeitos.

Builds motion for product UI on the web. It decides whether something should move at all, then why, which tool (CSS transition, WAAPI, Motion, spring), which properties, curve, duration, interruption, exit and reduced-motion path, and it writes the code. Also covers stagger and choreography, gestures (drag, flick, sheets), motion tokens and animation performance.

Design, build, animate, and review professional presentation decks (PPT/PPTX). Use when creating or redesigning a slide deck where content structure, visual style, layout variety, charts, imagery, animation, Morph transitions, motion narrative, or overall polish matters; when turning raw material into a designed argument; when adding or fixing slide animation and transitions; or when auditing an existing deck's stati…

Picks the right library for a frontend task from a curated, status-checked list, instead of hand-rolling it or installing an abandoned package. Use it when the user asks "what should I use for…", "is there a library for…", or names a package, and when you are about to build a toast, command menu, drawer, date picker, table, virtual list, drag and drop, chart, OTP input, animated number, theme switch, form validation…

Create and refine fast, clean Remotion motion graphics and visual explainers with strong opening hooks, restrained visuals, sourced icons and logos, ElevenLabs narration, and precise frame-based animation. Use for code-generated videos, not interactive website animation.

Makes AI-built UI look designed instead of generated. Shows three real design directions side by side, lets the user pick one, saves it as a DESIGN.md the agent follows on every screen, then rebuilds or polishes the UI pass by pass and checks it for AI tells.

Measures and improves Flutter mobile startup, frame rendering, scrolling, animation, memory, battery, network performance, and service reliability using repeatable evidence. Use for slow or janky apps, performance budgets, profiling, or production reliability.

Take a plain, sloppy or "AI-template" website (portfolio, personal site, landing page, small product site) and redesign it, phase by phase with the owner, into a clear, attention-grabbing, well-crafted, responsive site with its own identity.

Use when creating or revising academic research presentations, full-slide images, standalone scientific illustrations, or editable slide recreations.

演示配图：把方案、汇报、立项材料和文档里的内容做成 16:9 的图片，每页是一个 1920×1080 的 HTML 页面，用无头浏览器渲染成 3840×2160 的 PNG。包括 27 种常见版式的共用样例页（封面、目录、要点、数字、表格、各类图表、时间轴、流程、团队、代码、结束页等）、所有风格通用的组件和文字规范、可以继续增加的多种风格，以及创建、渲染、检查、导出和截取帖子的 TypeScript 脚本。当用户需要给文档做配图、做幻灯片那样的图，改、补、检查、导出这类配图，或者要把一套新的配图风格加进来时使用。不负责 .pptx 文件、在线幻灯片、海报和文档正文。

用于把 AI 生成或现有的软件界面图片整理为可与开发协作的设计稿：分析可见页面结构，拆分或重建界面元素，按需写入 Figma，并记录可核验的设计信息和待确认规则。也支持海报、商品图、插画、照片等通用图像分层；不从静态图推断未展示的业务行为。

设计、实现、审查和优化前端动效。当用户要求做动画、加动效、页面转场、 滚动动画、hover 反馈、加载动画、粒子或 3D 效果、视差、骨架屏过渡、 数字滚动，或提到 Framer Motion、motion、GSAP、ScrollTrigger、Three.js、 React Three Fiber、Lottie、View Transitions、Web Animations API、 CSS transition/animation、spring、easing 时使用。 触发词：动画、动效、转场、过渡、入场、出场、滚动效果、视差、 微交互、弹性、缓动、粒子、3D、animation、motion、transition、 animate、easing、spring、stagger、parallax、keyframes。

Create, recreate, improve, render, and evaluate animations from a written description, an image, a video, or an accessible reference link. Use for motion design, animated logos and icons, loaders, UI transitions, kinetic typography, explainers, procedural 2D/3D scenes, character or cinematic motion, and interactive animation.

用 Figma 设计稿开发 Web 前端 UI 时使用，解决「直接照抄画布数据导致还原度差」的问题。核心是「双通道阅读 + 验证闭环」：同时调用 get_figma_data（拿精确坐标/色值/字号/间距）和 download_figma_images（导出节点图片让模型"看见"布局与文案），禁止只用单一通道就写代码；实现后用 Playwright 截图对比设计稿自查、人工挑刺再迭代。适用于接入 Framelink figma-developer-mcp（工具为 get_figma_data / download_figma_images）的 Web 前端环境（React/Vue/HTML 等）。触发场景：用户给 Figma 链接/fileKey/nodeId 让实现 Web 页面、抱怨 Figma 还原效果差、或要求"按设计稿精确还原页面/组件"。

Use when adding or changing anything a user sees or clicks in a Vue 3 or React web app — a screen, component, composable or hook, store, form, route or data fetch — even for a small tweak or when the request only says "UI", "page", "form", "frontend" or "the button does nothing".

用 Figma 设计稿开发桌面端 UI（Qt Widget / QML / WinUI / Flutter 等非 Web 框架）时使用，解决「直接照抄画布数据导致还原度差」的问题。核心是「框架无关的数据→目标框架翻译方法论」：同时调用 get_figma_data（精确坐标/色值/字号/间距）和 download_figma_images（视觉截图）双通道阅读、禁止单通道写码、用确切值不做视觉估算、结构分析后按「画布概念→目标框架」映射表翻译（布局/颜色/间距/圆角/阴影等）、实现后实际运行程序截图对比设计稿自查 + 人工挑刺迭代。适用于接入 Framelink figma-developer-mcp（工具为 get_figma_data / download_figma_images）的桌面端环境。

Make a logo-led brand identity film in Remotion for any brand, in the language of OpenAI's "Refreshed" film. A white canvas with ink, the logo's own dot (or one signature element) leading every move, the mark drawn from its own construction lines and measurements, the logo zooming out into the brand's real website, the wordmark spelled letter by letter on the music, real product UI driven by a real mouse with measure…

Plan, art-direct, implement, or review distinctive motion-rich and 3D web experiences when immersion materially supports the brief. Use for cinematic heroes, scroll narratives, WebGL, interactive portfolios, premium launches, or explicit immersive frontend requests; do not invoke for ordinary UI work or as a replacement for the B16 Dev OS workflow.

Stress-tests a piece of UI with worst-case data and every reachable state, then reports what broke with a numbered fix for each. Use it when the user types /break, or asks to "break this", "stress-test", "try the worst case", "what happens with long names", "check the empty and error states" or "find the edge cases" on a component or screen.

Builds distinctive, production-grade, accessible frontend UI that does not look AI-generated. Use this skill whenever the task produces or changes anything a person sees in a browser - building or restyling a component, page, landing page, pricing page, dashboard, form, login or signup screen, settings panel, navbar, modal, table, email-style HTML, design system, theme, or design tokens - and for any "make it look be…

Design and build clean, polished, production-grade UI/UX — for new builds and redesigns — using design tokens, component specs, motion rules, and anti-AI-look taste guidance. Use when the user asks to design, build, restyle, modernize, or polish a user interface, landing page, dashboard, app screen, design system, or component library.

Reusable Xcrawl skill definitions for multi-agent runtimes, focused on API-first web data workflows.