Video & Animation

Plan, edit, grade, subtitle, and quality-check short-form vertical video (vlog, campus/travel, interview, brand story). Use when the user asks to turn raw clips into a 45–90s vertical vlog, add burned-in Chinese subtitles, build title/fact/end cards, grade flat/Log footage, deliver a finished MP4, or revise an already-approved cut without regressing it. Triggers include: 剪 vlog, 竖屏视频, 加字幕, 出成片, 换素材, 调色, 修改视频, 视频复盘, 已批准版本, 保留原版. Do NOT use for long-form film/TV finishing, pure audio editing, 3D animation, or when the user only wants a text plan.

Builds one finished piece of interface that moves well from a plain description, in a look dealt fresh for each request. It makes a control such as a switch or a slider, an animated chart, a widget or card, a screen or page inside someone's app, or a short looping animation that demos an app. It writes the piece into the person's project in its own framework, or as one self-contained HTML file when there is no project, and adds motion to a piece the person already has. Use it whenever someone wants such a piece to look and feel good, even in a beginner's loose words like "make it pop" or "feels nice to click", and even when they never mention motion. Whole websites and landing pages belong to page-design skills, and settle builds only the moving piece inside one; video editing and title sequences belong to video tools, and reviews of existing animation code to animation-review skills.

Make an animated video in a finished visual style (cut-paper, 80s anime, cartoon, collage, diagrams, 3D and more: 3 styles), animated in code and rendered with HyperFrames. Use when the user asks for a video, a film, a Short, an explainer or a ReelStyles style.

Transcribe speech locally with word-level timestamps (faster-whisper, no API key) and burn styled, word-highlighted captions into a video, or export SRT. Use when the user wants subtitles or captions on a video, a transcript with word timings, or cut points that never land inside a word.

Entry point for editing videos with ffmpeg from the terminal, especially short-form vertical video (Shorts, TikTok, Reels) made from screen recordings, app captures, gameplay or music content. Use when the user asks to edit, cut, caption, polish, loop, QC or publish a video and you need to pick the right skill and order of steps. Routes to auto-cut, hook-overlay, captions, audio-polish, loops-gifs-stickers, render-qc and youtube-publish.

Use when the user runs /reel-split, or asks for the split-screen explainer Reel — an animated widget panel on the top ~45 % of the frame, the speaker below it the whole time, no subtitles, no SFX — typical for "I found this tool / here are 5 tips" videos, on a talking-head take in raw/.

Routes design and front-end work to the right expert playbook from a bundled library of 328. Use for any UI, visual design, interaction, animation, typography, color, accessibility, layout, or copy work; for landing pages and design-engineering builds; for user research, personas, journey maps, usability tests, design systems, tokens, critique, and handoff; for Three.js and 3D scenes, motion showreels, video presentations, and SVG illustration; and for game architecture, combat, and VFX. Also use when asked to review, critique, or improve an existing interface, or when the request is "make this look better", "this looks AI-generated", "give it taste", or "which skill covers X".

Build frame-accurate, repeatable videos with React and Remotion, including compositions, transitions, animated text, media, audio, captions, and render validation.

Makes short films where every frame and every sound is computed by code in one HTML page, with no images, footage or samples. Covers any kind (explainers of technical systems, music videos, story films, ads, visual poems) in any hand-made look (riso, paper collage, ink, watercolour, pixel, blueprint, sand). Starts from a written intent card, judges against a gallery of 29 code-drawn films people loved, and verifies the finished MP4 against that intent. Ships a frame engine with a riso press and moves, a computed score, a one-command build to MP4, and checks for defects, joins, look and sound. Also pulls apart a reference clip, still or HTML page. Use when the user wants an animated explainer video, a code-drawn film, music video or still, a riso or print look, or to match or learn from a reference film. Not for a narrated lesson for YouTube, which belongs to lesson-film, or an animated diagram inside a page or doc.

Product demos, store previews, App Store header/search videos and loops cut from the user's REAL screen recording: trimmed, sped, zoomed, framed, captioned, voiced with local ffmpeg. Models add only what is not the product: narration (Eleven v3, Gemini TTS for Turkish), music and UI sound. Use when: "make a demo video of my app", "App Store preview video", "App Store header video", "Google Play promo video", "turn this screen recording into a demo", "GIF loop for my changelog", "Product Hunt launch video", "ekran kaydımdan tanıtım videosu yap", "changelog için GIF döngüsü". Chain: takes recordings, device frames (`layers`) and end cards from `launch-graphics`; produces MP4, WebM, GIF, SRT/VTT and posters. NOT for: generated b-roll shorts (use `short-video`); explainers with generated pictures (use `video-explainer`); paid ads (use `ads`); store screenshots, link cards (use `launch-graphics`); one generated clip (use `video`); voice or music alone (use `audio`); joining finished files (use `pipelines`).

根据主题和已录制视频完成剪辑，用 Remotion + FFmpeg 制作口播同步字幕、关键词高亮、章节标题、动态卡片、数据与截图动画，交付完整成片、封面及发布文案。也支持从网页、产品素材或脚本生成连贯中文配音视频，复用本人授权声音，处理口播节奏，以及 4:3、3:4、16:9 视频和封面。

Build Douyin/TikTok-style English sentence-pattern short videos using Remotion, Codex, and Alibaba DashScope CosyVoice TTS. Use for English sentence skeletons, speaking patterns, writing upgrade sentences, emotional expressions, grammar-in-context, and high-retention sentence drills with black/yellow high-contrast visuals, audio-driven subtitles, slot replacement animations, and Douyin-style hooks.

Build an interactive isometric "anatomy" of a real machine or invention (a lamp, a cipher machine, a steam engine, a transistor) from a pure tested model, with every caption fact tied to a source, a choice of colour theme and render style (line, dither, halftone, hatching, pixel), physically sensible motion, an exploded view, and a recorded 30 s film with soft sound in 16:9, 9:16 or square. Use for "anatomy of <machine>", "how does <machine> work, drawn", "isometric explainer", "exploded view video", "làm anatomy", "dựng máy <X>", or any request to dissect a real machine as an interactive figure or short film.

Claude Video 2.0. Make studio-quality motion graphics with Claude Code, free. It opens a gallery of 21 looks playing as real videos so the person picks one, asks 5 short questions, then builds the piece in code (HTML + GSAP, rendered to MP4 with HyperFrames). Pure graphics, or with the person's own photos and clips, or with AI clips from Higgsfield behind a storyboard and credit check that waits for "yes". Every piece is checked from real frames: 3 stills to approve, a pacing check, and a pixel loop check. Use when someone says "make me a motion graphic", "make a video intro", "animate this", "make a loop", "make a title card", "motion graphics with my photos", "turn my photo into a clip", "Claude Video", or "/claude-video-studio".

Thumbnails and covers that survive at small size: YouTube thumbnails (16:9), Shorts, Reels and TikTok covers (9:16), podcast and course covers (1:1), from a title plus an optional face photo, @handle character, brand colours or a frame of the user's own video. Text baked in by GPT Image 2.5; faces locked through reference edits. Use when: "make a thumbnail for my video", "YouTube thumbnail with my face", "Shorts cover", "podcast cover art", "thumbnail from this frame", "change only the text on the thumbnail", "videoma thumbnail yap", "kapak görseli hazırla", "küçük resimdeki yazıyı değiştir". Chain: consumes an @handle from `character` and stills or frames from `image` and `video`; produces sized PNG and JPEG. NOT for: posters, photo edits (use `image`); logos (use `brandkit`); link cards, OG images (use `launch-graphics`); ad creatives (use `ads`); the video (use `video`, `short-video`); new characters (use `character`); batches (use `pipelines`); model choice (use `models`); setup problems (use `setup`).

Use when asked to make, change, shorten, retime or re-voice a video with voiceover for any app or project, such as an app guide, tutorial, demo or walkthrough, an announcement or explainer built from a storyboard image, or a promo, ad, reel or skit (Remotion, ElevenLabs).

Build ~55–70-second vertical Douyin bilingual reading shorts for 高考 / 四级 (CET-4) / 六级 (CET-6) students in the 双语新闻阅读 Caption News template — a captions-first social-news explainer of one real, sourced topic (sleep, screens, exercise, climate) written as a short Gaokao-style passage, big white English caption chunks landing phrase by phrase with one keyword on the yellow block and Chinese under each, simple animated data graphics, SOURCE tags on every number, a reading question with 原文定位 and a key-word recap, with the locked 小猪 brand pig. Use when `npm run plan` names reading-caption-news or the owner asks for a news / science reading episode.

Upload finished videos or Shorts to YouTube safely from the terminal. Use when the user wants to authorize a YouTube channel, upload one or many videos with titles, descriptions and tags, schedule a publish time, or make sure a batch never double-posts. Defaults to private uploads and keeps an upload ledger.

Use when the user runs /reel-setup, has just cloned Arta Edit Studio, or a reel command fails because Python packages, Node packages, ffmpeg, fonts or reel.config.json are missing.

Post to Instagram and schedule Instagram posts, Reels and carousels from OpenClaw, Hermes, Claude Code or Codex. Publishes to the user's own Instagram account via the 1Social MCP server (OAuth, no API key), checks the Instagram caption, hashtag, aspect-ratio and media rules first, schedules up to 30 days ahead and returns a delivery receipt. Use to post or schedule on Instagram, publish a photo, Reel or carousel, plan Instagram content, or check, edit or delete Instagram posts.

Comprehensive Master Guide for Programmatic Video Creation in React with Remotion. Merancang, merakit, menganimasikan, menyinkronkan audio narasi/VO, dan merender video motion graphic profesional (16:9 explainer, 9:16 vertical shorts/TikTok/Reels) secara terprogram menggunakan React, Remotion 4.x, Spring physics, GSAP, Tailwind/CSS, dan CLI hardware-accelerated rendering. Gunakan skill ini setiap kali user meminta membuat video berbasis Remotion, merender MP4, membuat kinetic typography, transisi adegan (Series/TransitionSeries), sinkronisasi voiceover, atau optimasi render performance.

Make a motion-graphics music video (2D flat characters and kinetic lyrics, no filmed footage) from a song file. First builds a "song folder" with everything a model cannot hear (beats, bars, energy sections, drops, per-frame loudness per stem, kick/snare/hat hits, optional notes, timed lyrics), then plans a frame-deterministic 2D dance video on the beat and optionally renders it with HyperFrames. Load before planning, building or reviewing a music video, lyric video or beat-synced animation, or when the user says "make a music video for this song", "sync this animation to the beat", "analyse this song".

Ads from the user's REAL product: UGC and polished ad videos (presenter, unboxing, showcase, testimonial, TV spot) and static ads with copy. Physical products: keyframes locked to the product photo, Seedance clips or TTS + OmniHuman, GPT Image 2.5 copy. Apps and SaaS: real captures and recordings, composited, never regenerated. Use when: "make a UGC ad for this product", "15 second TikTok ad", "static ad with this headline", "app install ad from my screen recording", "A/B ad variants", "bu ürün için reklam videosu yap", "uygulamam için reklam hazırla", "indirim görseli hazırla". Chain: consumes `product-photoshoot` stills, `character` @handles, `launch-graphics` pages; produces ads and a manifest. NOT for: text-free product photos (use `product-photoshoot`); posters (use `image`); listing cards (use `marketplace-cards`); store screenshots, OG cards (use `launch-graphics`); one clip (use `video`); organic shorts (use `short-video`); product demos (use `demo-video`); thumbnails (use `youtube-thumbnail`).

将一个 GitHub 仓库从工作台入库、官方来源与高风险声明核验、抖音内容策划、真实网页证据采集、脚本与 Timeline、Remotion 渲染、配音、两尺寸封面到五文件发布包完成闭环。用户要求介绍 GitHub 单仓库、复制现有 GitHub 视频任务、录屏或批量生产 GitHub 项目视频时使用。