Media Skills

Generate publication-quality academic illustrations through a local Codex app-server bridge that uses Codex native image generation. This is a separate experimental alternative to `paper-illustration`, intended for Claude Code users who want a GPT-image-style renderer without modifying the original skill.

Generate publication-quality AI illustrations for academic papers using Gemini image generation. Creates architecture diagrams, method illustrations with Codex-supervised iterative refinement loop. Use when user says \"生成图表\", \"画架构图\", \"AI绘图\", \"paper illustration\", \"generate diagram\", or needs visual figures for papers.

Generate pixel art SVG illustrations for READMEs, docs, or slides. Use when user says \"\u753b\u50cf\u7d20\u56fe\", \"pixel art\", \"make an SVG illustration\", \"README hero image\", or wants a cute visual.

Run comprehensive UI tests for the Image Studio feature. Covers Media Library entry points, Edit Mode, Generate Mode, Block Editor integration, navigation, and delete. Use when running the full UI smoke test or testing any Image Studio surface.

GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora.

PyTorch library for audio generation including text-to-music (MusicGen) and text-to-sound (AudioGen). Use when you need to generate music from text descriptions, create sound effects, or perform melody-conditioned music generation.

Vision-language pre-training framework bridging frozen image encoders and LLMs. Use when you need image captioning, visual question answering, image-text retrieval, or multimodal chat with state-of-the-art zero-shot performance.

Large Language and Vision Assistant. Enables visual instruction tuning and image-based conversations. Combines CLIP vision encoder with Vicuna/LLaMA language models. Supports multi-turn image chat, visual question answering, and instruction following. Use for vision-language chatbots or image understanding tasks. Best for conversational image analysis.

Foundation model for image segmentation with zero-shot transfer. Use when you need to segment any object in images using points, boxes, or masks as prompts, or automatically generate all object masks in an image.

State-of-the-art text-to-image generation with Stable Diffusion models via HuggingFace Diffusers. Use when generating images from text prompts, performing image-to-image translation, inpainting, or building custom diffusion pipelines.

OpenAI's general-purpose speech recognition model. Supports 99 languages, transcription, translation to English, and language identification. Six model sizes from tiny (39M params) to large (1550M params). Use for speech-to-text, podcast transcription, or multilingual audio processing. Best for robust, multilingual ASR.

Build and publish a Gradio demo on Hugging Face Spaces for a user-provided LoRA. Use when someone asks to create, generate, ship, or publish a Space, demo, Gradio app, or playground for a LoRA — including LoRAs for Qwen-Image, Qwen-Image-Edit, LTX-Video, Wan, FLUX, SDXL, or other diffusion base models. Also triggers when someone describes a LoRA they trained or hosts on the Hub and wants to share it. Covers picking the right base pipeline and `diffusers` inference recipe, designing a UI tailored to the LoRA's task and inputs (Union/multi-task control, edit, video, image, etc.), respecting model-card recommendations (trigger words, steps, guidance, LoRA scale, example inputs), and shipping to ZeroGPU hardware as a private Space by default.

Use Transformers.js to run state-of-the-art machine learning models directly in JavaScript/TypeScript. Supports NLP (text classification, translation, summarization), computer vision (image classification, object detection), audio (speech recognition, audio classification), and multimodal tasks. Works in browsers and server-side runtimes (Node.js, Bun, Deno) with WebGPU/WASM using pre-trained models from Hugging Face Hub.

Create, inspect, and update Obsidian JSON Canvas boards with text, file, link, group, and edge nodes. Use for canvas status, canvas lists, visual maps, zones, spatial layouts, adding vault notes or media to a .canvas file, and requests such as create canvas, add to canvas, or put this on the canvas.

把一篇文章或口播稿，做成"看起来像视频"的点击驱动 16:9 网页演示，可选合成口播音频。流程：原始文章 → **一次产出**口播稿 + outline 开发计划 → 用户**一次对齐** 5 件事（稿子 / outline / 主题 / 素材 / 开发模式）→ 网页开发（逐章 / 顺序 / 并行）→ 可选音频合成（provider-agnostic：内置 MiniMax mmx-cli + OpenAI TTS，可换 ElevenLabs / edge-tts / Azure / 自带 TTS）。**outline 只规划节奏与信息密度，不规划动画** —— 动画由章节开发时按 PRINCIPLES + ANTI-AI 法则即时设计。每次点击推进口播稿的一个节拍，每一步独占整屏，进度条平时隐藏只在悬浮时出现。适用场景：用网页做视频（动态 PPT 但不像 PPT）、把口播稿 / 文章变成可交互的解说、为 B 站 / YouTube / 视频号录屏教程、做有电影感的产品 / talk demo。本 Skill 沉淀的是设计方法论 + 协作流程 —— 不绑定任何特定样式 / 字体 / 颜色 —— 因此能复用到任意主题与美学。

面向 GPT Image 2 的图像生成 / 编辑技能。可在 3 种环境下使用：(A) Garden 本地模式，通过 OpenAI 兼容接口直接出图并落盘；(B) Host-Native 模式，把本 Skill 当作提示词工程指引，把渲染好的 prompt 交给宿主 Agent 自带的图像工具出图；(C) Advisor 模式，宿主无任何图像工具时退化为高质量 prompt 顾问。涵盖 18 大类、80+ 个结构化模板，覆盖海报 / UI / 产品 / 信息图 / 学术图 / 技术架构图 / 漫画 / 头像 / 流程板 / 电影分镜 / IP 周边 / 编辑工作流等场景。

Performs a strict evaluation of a video asset using Google's official 'ABCD' framework (Attract, Brand, Connect, Direct) based on transcript and metadata.

Provides a high-signal briefing on events in a specific location and timeframe, backed by primary video sources and transcripts.

Deconstructs high-performing or viral videos to extract actionable creative insights from metadata and transcript.

Performs high-density targeted extraction across 10+ videos to map semantic landscapes, consensus, and controversies.

Launch and supervise OpenShell gator agents. Use when starting gator on issues or PRs, checking gator sandboxes, building the gator sandbox image, restarting stuck gators, inspecting gator logs, or experimenting with gator harness/model overrides. Trigger keywords - launch gator, start gator, run gator, gator sandbox, supervised gator, gator logs, restart gator.

Generate walkthrough videos from Stitch projects using Remotion with smooth transitions, zooming, and text overlays

>- Upload local assets (images, mockups, extracted HTML, design markdown) to a Stitch project. ALWAYS use this skill when you need to upload visual assets, HTML pages, or design docs to Stitch, particularly when direct MCP tool calls fail or truncate due to base64 token limits.

Audit paid-ad copy, images, video, hooks, concepts, format coverage, platform-native fit, message match, creative fatigue, accessibility, and policy across supported platforms. Use for creative audit, ad creative, creative fatigue, creative diversity, ad copy review, video review, image review, or production priorities.

Extract a public-safe brand and offer profile for paid advertising from an authorized website and operator input. Triggers on: brand DNA, brand profile, brand identity, brand style, brand colors, brand voice, visual identity, style guide, website brand analysis.

Generate paid-ad image assets from a validated creative brief and brand profile using an explicitly configured image provider. Triggers on: generate ads, generate ad images, create ad creatives, create ad images, make ad images, generate visuals, make campaign visuals, generate images from campaign brief.

Generate rights-cleared paid-ad product photography variants from an authorized source image and validated brand profile. Triggers on: product photo, product photography, virtual photoshoot, photoshoot, enhance product image, studio shot, lifestyle shot, lifestyle product image, floating product image, ingredient shot.

Audit YouTube Ads campaign setup, video and Demand Gen inventory, Shorts, in-stream, CTV, creative, audiences, brand safety, bidding, and measurement. Use for YouTube Ads, video ads, pre-roll, bumper ads, skippable in-stream, Shorts ads, Demand Gen, VAC migration, CTV, or YouTube campaign optimization.

>- Skill to run xcodebuild tests on iOS Simulator while recording a video walkthrough, dynamically selecting the highest available OS and device. Most useful for running XCUITests.

Guide teams through Lean UX Canvas v2. Use when framing a business problem, surfacing assumptions, and defining what to learn next.

Guide teams through MITRE's Problem Framing Canvas. Use when you need a clearer problem statement before jumping to solutions.

Evaluate an AI product idea across outcomes, hypotheses, risks, and positioning. Use when deciding whether an AI solution deserves investment or recommendation.

Prioritize stakeholders using two complementary grids. Use when setting engagement strategy and surfacing whose voice needs elevating after stakeholder identification.

Mine public reviews, app stores, and forums for unmet needs, competitor weaknesses, and switching triggers — with quoted evidence. Use when you want customer voice without waiting on interviews.

Updates Docker container image tags used by Aspire hosting integrations. Queries registries for newer tags, uses LLM to determine version-compatible updates, and applies changes. Use this when asked to update container image versions.

Create and publish concise Rivet launch or changelog posts through an approval-gated workflow, including release research, a required user-selected hero variation, local HTML-rendered social and technical images, an R2 hero upload, MDX authoring, a reviewed draft GitHub PR for manual merge, and a scheduled and verified Buffer thread for @rivet_dev. Use when asked to launch, announce, ship, or prepare and schedule a Rivet feature or release announcement.

This skill encodes Emil Kowalski's philosophy on UI polish, component design, animation decisions, and the invisible details that make software feel great.

Design and implement web animations that feel natural and purposeful. Use this skill proactively whenever the user asks questions about animations, motion, easing, timing, duration, springs, transitions, or animation performance. This includes questions about how to animate specific UI elements, which easing to use, animation best practices, or accessibility considerations for motion. Triggers on: easing, ease-out, ease-in, ease-in-out, cubic-bezier, bounce, spring physics, keyframes, transform, opacity, fade, slide, scale, hover effects, microinteractions, Framer Motion, React Spring, GSAP, CSS transitions, entrance/exit animations, page transitions, stagger, will-change, GPU acceleration, prefers-reduced-motion, modal/dropdown/tooltip/popover/drawer animations, gesture animations, drag interactions, button press feel, feels janky, make it smooth.

Create and present web-based slidedecks for developers using Slidev with Markdown, Vue components, code highlighting, animations, and interactive features. Use when building technical presentations, conference talks, code walkthroughs, teaching materials, or developer decks.

Iteratively author and test a custom Azure AI Content Understanding analyzer for a folder of **document** files (PDFs, page images) of a single type. Walks layout extraction → schema authoring → validation → batch test → agent review → refine cycle using the typed ContentUnderstandingClient. Document modality only — audio, video, and image analyzers are planned for a later update. Use when the user wants to author a custom analyzer for invoices, contracts, forms, or any other single-type document set.

Read this before building, booting, or configuring an azldev image. Explains the azldev image commands (list, build, boot, test, customize) and the [images.<name>] config (kiwi definition, capabilities, tests, publish); the kiwi XML format itself is upstream KIWI NG. Triggers include image build, image boot, kiwi, container image, VM image, images.toml.

Fetches complete Airbnb listing details for a given numeric listing ID via the internal GraphQL API, returning title, room type, description, amenities, photos, coordinates, city, house rules, highlights, ratings, review count, bedroom configuration, and property overview. Use when user mentions Airbnb listing details, Airbnb property info, Airbnb room details, get Airbnb listing data, Airbnb amenities list, Airbnb house rules, Airbnb property description, Airbnb detail page scraper, Airbnb rooms detail, Airbnb property page data, Airbnb listing info, fetch Airbnb room details, pull Airbnb listing.

Extracts full detail data from a single Goofish (闲鱼/xianyu, goofish.com) second-hand item page. Input: item URL or item ID. Output: title, price, seller info (name, labels), full description, image gallery, item tags/attributes, want-count. Use when user mentions goofish item detail, 闲鱼商品详情, xianyu item page, 二手商品详情, get goofish product info, 采集闲鱼单品数据, 抓取闲鱼商品, scrape goofish item, xianyu product detail, 获取闲鱼卖家信息, seller info goofish, item description goofish, 闲鱼详情页, 想要人数, 闲鱼图片. Also applies to: verifying a specific listing before purchase, extracting seller contact/rating information, bulk item detail enrichment from a list of item IDs.

Search Taobao and Tmall product listings by keyword, returning paginated product cards with title, price, shop, image, sales, and tags. Use when user asks to search Taobao, find products on Taobao/Tmall, scrape Taobao search results, get product listings from Taobao, collect Taobao items by keyword, 搜索淘宝, 淘宝关键词搜索, 采集淘宝商品, 抓取淘宝搜索结果, 淘宝天猫商品列表. Also applies to bulk keyword searches, price monitoring across keywords, and competitive product research on Taobao.

Fetch full product detail from a Taobao or Tmall product page by itemId, returning title, price, shop info, images, SKU variants, and product attributes. Use when user asks to get product details from Taobao, scrape a Taobao item page, extract product info by item ID, fetch Tmall product data, 抓取淘宝商品详情, 获取淘宝商品信息, 淘宝商品页面采集, 天猫商品详情, 按商品ID获取信息. Also applies to building product databases, price tracking by itemId, and product comparison research.

Fetch customer reviews for a Taobao or Tmall product by itemId, returning reviewer name, date, purchased variant, review text, and photo URLs. Use when user asks to get product reviews from Taobao, scrape Taobao customer feedback, extract buyer reviews by item ID, collect Tmall ratings and comments, 采集淘宝商品评价, 抓取淘宝买家评论, 获取淘宝商品评论, 天猫商品评价抓取, 按商品ID获取评价. Also applies to sentiment analysis of product reviews, building review datasets, and monitoring product rating changes.

This skill helps users automatically extract structured image data from Google Images via BrowserAct API. Agent should proactively apply this skill when users express needs like finding images for specific keywords, gathering product style images for competitors, building visual datasets at scale, scanning visual search results for market research, tracking localized image trends by country, compiling related image thumbnails and links, extracting image titles and source logos, fetching click through URLs from image results, monitoring competitor visual assets, sourcing creative content for specific topics, looking up product pictures in different regions, collecting structured image metadata without opening detail pages.

Scrapes tweets from an X (Twitter) user profile timeline given a handle, with selectable mode: tweets, tweets+replies, or media-only. Returns normalized per-tweet data including text, author profile, engagement counts, media, hashtags, mentions, and cursor for pagination. Use when user mentions X user tweets, Twitter user tweets, profile timeline, user feed, @handle tweets, scrape Twitter user, scrape X account, get all tweets from a user, scrape user timeline, download user posts, export tweets from user, with_replies tab, replies tab Twitter, media tab Twitter, photos of user Twitter, videos of user Twitter, latest tweets from user, recent tweets from account, KOL tweets, influencer tweet history, account post extraction, twitter handle scraper, x handle scraper, get tweets by username, monitor a Twitter account, monitor an X account, daily tweet export from handle, scrape from:username, twitter profile posts, x profile posts, by handle, by username. Also applies to building a creator backlog, tracking competitor accounts, populating quote-tweet research datasets, or any paginated bulk collection driven by a single user handle.