设计与三维建模

Animates Arabic text correctly and beautifully in Remotion, HTML/CSS/GSAP and After Effects-style workflows. Covers right-to-left layout, word-level (never letter-level) animation, Arabic fonts, mixed Arabic/English, digits, diacritics clipping, and kinetic typography synced to speech. Use when making Arabic titles, kinetic text, lower thirds, animated captions or any motion graphic with Arabic text, or when Arabic letters appear disconnected, reversed, or cut off in an animation.

Writes complete Arabic video scripts (YouTube long-form, explainers, tutorials, talking-head, faceless) that keep viewers watching, in the creator's dialect, with timing based on Arabic speech rate and a visual cue for every beat. Use when the user wants "سكريبت", "اكتبلي فيديو", "سيناريو يوتيوب", an Arabic YouTube script, or to turn an article/idea into an Arabic video.

Writes complete Arabic video scripts (YouTube long-form, explainers, tutorials, talking-head, faceless) that keep viewers watching, in the creator's dialect, with timing based on Arabic speech rate and a visual cue for every beat. Use when the user wants "سكريبت", "اكتبلي فيديو", "سيناريو يوتيوب", an Arabic YouTube script, or to turn an article/idea into an Arabic video.

Plans and writes Arabic short-form videos (Reels, TikTok, YouTube Shorts, Snapchat Spotlight) end-to-end in one pass, covering hook, 20–60 s script, shot list, on-screen text, caption and hashtags, tuned for Arab audiences and RTL screens. Use when the user asks for "ريل", "ريلز", "تيك توك", "شورتس", "فيديو قصير", or a quick vertical video in Arabic.

Plans and writes Arabic short-form videos (Reels, TikTok, YouTube Shorts, Snapchat Spotlight) end-to-end in one pass, covering hook, 20–60 s script, shot list, on-screen text, caption and hashtags, tuned for Arab audiences and RTL screens. Use when the user asks for "ريل", "ريلز", "تيك توك", "شورتس", "فيديو قصير", or a quick vertical video in Arabic.

Creates Arabic YouTube titles and thumbnail text as a matched pair, handling right-to-left layout, mixed Arabic/English, digit style, length limits, and readability at phone size. Use when the user asks for "عنوان", "عناوين يوتيوب", "نص الثمبنيل", "تصميم الصورة المصغرة", or Arabic titles/thumbnails for a video.

Creates Arabic YouTube titles and thumbnail text as a matched pair, handling right-to-left layout, mixed Arabic/English, digit style, length limits, and readability at phone size. Use when the user asks for "عنوان", "عناوين يوتيوب", "نص الثمبنيل", "تصميم الصورة المصغرة", or Arabic titles/thumbnails for a video.

AI 智能体架构师：设计智能体系统：单智能体与多智能体、工具与 MCP、记忆、防护栏、评估。

智能合约系统架构师：设计合约系统的模块、可升级性、访问控制、存储和审计就绪能力。

DeFi 协议架构：为 AMM、借贷、金库和衍生品设计机制：预言机、清算、攻击面。

架构决策框架。需求分析、权衡评估、ADR文档。适用于做出架构决策或分析系统设计时。

掌握经过验证的后端架构模式，包括整洁架构、六边形架构和领域驱动设计，以构建可维护、可测试且可扩展的系统。

Record a scripted terminal demo of a CLI or TUI with asciinema: hidden fixture setup, commands typed on screen, TUI keys driven through tmux, a leak gate on the cast, scene-by-scene proof by replaying the cast, an agg GIF for the README, and upload to asciinema.org. Activates when the user says 'asciinema', 'record a terminal demo', 'screencast the CLI', 'record the TUI', 'add a demo GIF to the README', 'make a cast', or 'upload to asciinema'. Do not use for browser recordings (agent-browser) or produced motion video (video).

Use when optimizing images or other assets in Astro — <Image />, <Picture />, getImage(), Fonts API, or OG image generation.

Deterministic, non-generative image transform MCP server: reproducible recipes, immutable originals.

Clean up or mix a video's audio track — normalize loudness, reduce background noise/hiss, or add background music (optionally ducked under speech). Use whenever the user wants to fix quiet/loud/inconsistent audio, remove hiss/hum, or lay music under a talking-head/voiceover. For removing um/uh filler words use filler-removal instead; for replacing the audio entirely use plain ffmpeg map.

Designs how a system knows who someone is and what they may do — login, sessions, tokens, roles, and permission checks. Use this whenever the user is building login or signup, adding roles or permissions, choosing between sessions and JWTs, integrating OAuth or SSO, or asking how to protect an endpoint. For finding authorization bugs in code that exists, use security-analysis.

使用 Zafiro.Avalonia 构建现代 Avalonia UI 布局的指南，强调共享样式、通用组件和避免 XAML 冗余。

Image, video, audio, face-swap, talking avatars and chat across 300+ AI models, one balance.

Design-review copilot: industry design checkpoints, edge cases, and Japanese legal notes.

Search and read bard0 design's FPGA and embedded engineering articles and site pages.

从无障碍性、布局、文案、排版、色彩和 UI 打磨等方面审查一个屏幕、流程或整个界面，并返回一份带排名的报告。

设置分组、对齐、阅读顺序、响应式结构以及译文所需空间，使布局在调整大小、翻译或镜像时依然稳固。

设定并审阅文本的渲染方式，从字号层级和间距到字体特性、换行、截断和标点。

Applies exact UI polish values: concentric border radius, optical alignment, layered surface shadows, neutral image outlines, a 0.96 press scale, icon cross-fades, and theme-switch transition suppression. Activates when editing component CSS or Tailwind, when an interface looks slightly off, or when the user says 'polish the UI', 'better UI', 'concentric radius', 'optical alignment', 'shadow instead of border', or 'icon transition'. Do not use for design-system choice, accessibility audits, or building the page - that is vd:uiuxdesign. Do not use for posting a PR review - that is vd:code-review.

以精确数值打磨界面表面、图标和动效，涵盖圆角半径、光学对齐、阴影、图标状态和动画。

让智能体在你自己的 Windows 桌面上驱动 Revit、AutoCAD、SketchUp、Blender 和 PDF 图纸集。

在检查 Blender 角色、绑定、姿势、动画重定向、地面接触、朝向，或仅凭截图不足以判断的模型与动作对齐时使用此技能。

Design and run a head-to-head comparison whose result you can actually believe — blinding, balanced position assignment, a control that could plausibly have won, and honest reading of a small sample. Use this whenever you are about to claim one approach beats another: A/B-ing two prompts or configs, comparing a new pipeline against the old one, benchmarking a library or model against an alternative, testing whether a change actually helped, or judging which of two outputs is better. Also use it when someone hands you a comparison result and asks whether it holds up, when a result looks suspiciously clean, or when you notice you are about to declare a winner after looking at the outputs unblinded. Not for checking that your own work is finished (use verification-before-completion) and not for wording the claim once evidence is in (use falsifiable-claims).

当用户需要 Claude 不具备的能力（图像生成、实时 X/Twitter 数据）或明确要求使用外部模型（"blockrun"、"use grok"、"use gpt"、"dall-e"、"deepseek"）时使用

系统架构师和技术设计负责人。适用于用户要求与 Winston 交谈或请求架构师。

系统架构师和技术设计负责人。适用于用户要求与 Winston 交谈或请求架构师。

运用多种创意技巧主持头脑风暴会议。适用于用户说“帮我头脑风暴”或“帮我构思”时。

运用多种创意技巧主持头脑风暴会议。适用于用户说“帮我头脑风暴”或“帮我构思”时。

以两份文档记录用户的 UX 愿景：DESIGN.md 描述产品外观，EXPERIENCE.md 描述产品行为。用户说“lets create UX design”“create UX specifications”或“help me plan the UX”时使用。

以两份文档记录用户的 UX 愿景：DESIGN.md 描述产品外观，EXPERIENCE.md 描述产品行为。用户说“lets create UX design”“create UX specifications”或“help me plan the UX”时使用。

Explore the solution space when the path isn't obvious - invent options, stress-test them, pick one. Use for architecture decisions, design tradeoffs, and when the user asks 'how should I approach X?'. Default produces a decision brief; pass `--quick` for chat-only, `--deep` for multi-round adversarial debate with full design doc. Do not use when want is unconfirmed (vd:interview), when the user says 'grill me' (vd:interview --grill), or when the deciding will not fit one session (vd:interview --wayfinder).

设计实验以验证现有产品的假设——原型、A/B 测试、技术探针及其他低成本验证方法。适用于验证假设、低成本测试功能想法或规划产品实验。

为新产品设计精益创业实验（预原型）。创建 XYZ 假设，并建议低成本的验证方法，如落地页、讲解视频和预购。当验证新产品创意、创建预原型或测试市场需求时使用。

与用户一起头脑风暴，生成、组合并完善非显而易见的想法，形成有前景的概念。适用于用户想要头脑风暴、构思、探索可能性、摆脱显而易见的方法或寻找创造性解决方案时。

在任何创意或构建工作之前你必须使用此技能：创建功能、构建组件、添加功能、修改行为，或规划任何新事物，无论软件内外（网站、演讲、业务）。一次一个问题地弄清用户真正想要什么以及为什么，比较2-3种方案，并在开始构建前获得设计文档批准。即使用户从未提到头脑风暴或设计也要使用；仅对完全明确的一行任务跳过。

Use before implementing any feature, behavior change, or refactor - settles requirements and design, then gets the design independently reviewed and approved by the user before code is written

在任何创造性工作之前必须使用此技能——创建功能、构建组件、添加功能、修改行为，或规划任何新事物，无论是否涉及软件（如演讲、创业、装修）。

Use before implementing any feature, behavior change, or refactor - settles requirements and design, then gets the design independently reviewed and approved by the user before code is written

将 Anthropic 官方品牌配色与字体应用于任何需要体现 Anthropic 视觉风格的产物。当涉及品牌配色或样式指南、视觉格式或公司设计标准时使用。

将 Anthropic 官方品牌配色与字体应用于任何需要体现 Anthropic 视觉风格的产物。当涉及品牌配色或样式指南、视觉格式或公司设计标准时使用。

将 Anthropic 官方品牌配色与字体应用于任何需要体现 Anthropic 视觉风格的产物。当涉及品牌配色或样式指南、视觉格式或公司设计标准时使用。

Reclaim disk space on this machine and keep it from filling again — run the reclaim report, decide the rows that need a human, and hunt for consumers no rule covers yet. Owns `claude/scripts/reclaim-disk.sh` (the deterministic sweeps and the low-space alarm) and the judgment the script cannot encode: classifying a consumer nobody has seen before, and deciding whether prevention beats sweeping. TRIGGER when: the volume is low or a command failed with ENOSPC; a build directory, cache or VM image has grown large; the low-space alarm fired; a new machine needs the alarm installed; or the user asks how to stop something filling the disk again. DO NOT TRIGGER when: the task is uninstalling an app and its data (that is `cleanup`), removing a Docker volume for a project (that is the project's own concern), or deleting a file the user named — just delete it.

使用浏览器自动化 MCP（claude-in-chrome、Playwright 或 Puppeteer）执行部署后自动化 UI 验证：控制台错误和 Core Web Vitals 冒烟检查、表单和认证流程交互测试、跨三个断点的截图视觉回归，以及以 SHIP / DO-NOT-SHIP 结论收尾的 axe-core 无障碍审计。当在预发布或预览环境测试已部署功能、发布前端变更前、审查前端 PR，或检查响应式布局和无障碍性时使用。

依据 Figma 文件或设计图片构建匹配设计的 UI，使用项目已有的设计变量和组件。