Video & Animation

Edit video and audio with local FFmpeg: cut, join, reframe, speed, captions, overlays, silence removal, sync, loudness, HDR to SDR, platform exports and checks. Use whenever the user mentions a video or audio file (mp4, mov, wav, m4a), footage, captions, a reel or short, YouTube/TikTok delivery, LUFS or ffmpeg, even without saying "edit". Needs FFmpeg 6.1+ and Python 3.10+; stdlib only, no cloud.

Design and build the visual layer of a long-form talking-head video as one persistent world the camera moves through, instead of a sequence of separate graphics. Use when asked to edit a video, add motion graphics, design the visuals for a script, plan an edit, or fix an edit judged confusing, text-heavy, hard to follow, or "clearly made by an AI". Covers the persistent world, camera grammar, the spotlight rule, spatial open loops, ornament and token discipline, transcript timing, and the verification gates.

Create editable TV news videos with code graphics and optional voiceover. Use for broadcast news, ticker videos, or news-style explainers. Do not use for avatar generation or footage editing.

Produce or improve Arabic software-training videos from an existing application, user guide, or narrated video. Covers real UI capture, Arabic voice-over and captions, progressive focus motion, original music, editable Python/FFmpeg production, and documented media QA. Use for narrated application walkthroughs; not for rebuilding an application or generic promotional videos.

Learns a video editing style from reference videos or a description, then edits your footage in it with an edit plan and review rounds. Triggers: editing style, style-lab, סגנון.

Turn ideas, markdown notes, code repositories, research papers, or documentation into 1080p animated explainer videos using the code-driven video engine. Universal skill for AI coding agents (Claude Code, Antigravity, Codex, DeepSeek Harness, OpenHands/OpenClaw, Cursor, etc.) to design storyboards, write page drawing animations in JavaScript, run layout checks, and render explainer videos.

Build a polished vertical promo or edit video for a music app, artist, album, playlist or song (Spotify-style ad, reel, TikTok/Instagram edit). It starts by asking for 3-4 style references that shape the result, then assembles the video in code from the user's logo, app screenshots, songs, sound effects and music-video clips. It covers official high-res cover art, kinetic typography, smooth matched zooms, beat-synced songs with seamless low-to-high handoffs, particle cover animations, 60/120 fps exports, QA and delivery. Use this skill whenever someone wants a video made from a logo, screenshots, album covers, songs or a music video, even when they don't say "promo", e.g. "make an ad with these Spotify screenshots and this song", "edit a reel for this track", "put this music video and cover into an animation", or when continuing or revising such a video.

21 Lessons, Get Started Building with Generative AI

Every plan of every GTM API on datacircle.dev, as one file you download and query. The file covers People, Contact, Companies, Signals, Social media, E-commerce, Real estate and Web. For each plan, the file lists the price per 1,000 results, our score for the plan and the rate limits. The file also lists the fields the plan returns and a link to the plan's docs. Use the file to compare or price a GTM API.

制作由需求决定时长的中文人物/主题短片；覆盖 brief、史实核查、旁白、节拍、段/切/关键帧、首帧、H3 投产包、评审、剪辑、配乐、终检与交付；30 秒和 3 分钟只是示例，不是硬范围。

AI配音、视频配音、文字转语音（TTS）、声音克隆与字幕生成技能：把口播稿、电影解说稿、小说推文、广告脚本、带货文案、有声书章节写成带 [角色] 标签的稿件，一条命令生成多角色配音 WAV、带时间轴的 SRT 字幕和逐句时间 manifest.json，直接导入剪映 / CapCut / Premiere / ffmpeg。音色可用账号已有角色，也可用参考音频 + 对应文本做声音克隆；支持语气标签、0.25–4 倍语速、停顿控制；已有音频 / 视频可识别成 SRT / TXT / JSON 字幕。内置口播、解说、带货、数字人、广告片、有声书 6 个节奏预设；改一句只重配一句。配音与字幕由 Miragea 开放 API 生成，本地只需 Python 3.9+ 标准库、无需显卡，可 --dry-run 离线试跑。适用于 Claude Code、Codex、OpenClaw、Cursor、WorkBuddy。AI voiceover, text to speech, voice cloning, multi-speaker dubbing and SRT subtitles.

Make a ~45-65 s vertical animated story short in a cartoon-in-3D style (flat 2D white-headed cartoon characters lit inside a colourful, ink-outlined 3D world; gliding camera with no hard cuts; word-by-word captions; fast narrator; music bed + SFX on every action), rendered with three.js inside Remotion on a Mac. Use when the user asks for a short "in that cartoon 3D style", "like the fake shoppers / grocery cart video", a history/science story short with cartoon characters in 3D, to suggest viral topic ideas for such shorts, research a topic's facts and write its script, or to measure / look-match against reference shorts. Built on mromidev/claude-animated-shorts (MIT) with every gap from its audit fixed.

小说推文配音、有声书配音、多角色配音技能：把小说推文文案、网文章节、故事、广播剧剧本按旁白和角色分配不同音色，一键生成整段配音 WAV、带角色名的 SRT 字幕和逐句时间轴 manifest.json，直接配画面做小说推文视频；段落、场景、章节停顿可控。适合小说推文号、故事号、网文作者有声化、睡前故事和广播剧小样。音色可用账号已有角色或参考音频声音克隆，固定旁白声线；语气标签和语速可调；改一句只重配一句。由 Miragea 开放 API 生成，本地只需 Python 3.9+ 标准库、无需显卡，可 --dry-run 离线试跑。适用于 Claude Code、Codex、OpenClaw、Cursor、WorkBuddy。Novel video voiceover, audiobook narration, multi-voice, SRT subtitles.

AI播客、播客配音、双人对谈播客技能：把文章、读书笔记、公众号文章或写好的对谈稿，整理成适合“听”的播客稿，单人主播或两位主持人各用一个音色，逐句配音并拼成整期节目 WAV，同时输出 SRT 字幕和逐句时间轴；再用 shownotes.py 根据稿件里的 `## 章节` 自动生成带时间戳的节目章节（chapters.txt / chapters.json）和完整文字稿 shownotes.md，可直接贴到小宇宙、喜马拉雅、Apple Podcasts 等平台的节目简介。改一句只重配一句。可用账号里的角色或本人授权的参考音频声音。由 Miragea 开放 API 生成语音，本地只需 Python 3.9+ 标准库、无需显卡，可 --dry-run 离线试跑；计费与限额以 capabilities 实时返回为准。适用于 Claude Code、Codex、OpenClaw、Cursor、WorkBuddy。AI podcast, podcast voiceover, two-host dialogue, show notes, chapters.

Record a mobile app bug with Maestro and turn it into a video people can follow: one run that shows the bug, or the bug build beside the fixed build. Use when: presenting a reproduced bug to a team or reviewer, attaching evidence to a bug report or pull request, or showing that a fix works. Triggers: bug video, before and after video, record the repro, show the bug, show the fix, demo the fix, present the bug.

3D with Three.js, React Three Fiber (R3F), Drei helpers, Rapier physics (@react-three/rapier) and postprocessing effects (@react-three/postprocessing). Use when building 3D scenes for web or video — models (GLTF), lights, cameras, environment maps, physics, bloom/vignette/depth-of-field. Inside Remotion, also follow remotion-markup/3d.md.

Estilo CARTAZ HISTÓRICO — mini-doc narrado com cerca de 80% da timeline coberta por animação editorial em Remotion, sincronizada palavra a palavra com a fala. Colagem de papel creme, laranja queimado e preto, pinturas e gravuras de época de fontes reais recriadas com fidelidade, nomes em grotesca que entram letra a letra, anos em blocos de cor, mapas, linhas do tempo, citações em máquina de escrever e cards de capítulo, entregue em camadas ProRes com alfa no Premiere e ligada ao After Effects por Dynamic Link. Use quando pedirem "estilo CARTAZ", "cartaz histórico", "o estilo do doc do liberalismo", "animar as falas do mini doc", "mini-doc animado", "documentário com colagem", "cobrir 80% com animação", "card de capítulo", "aviso inicial estilo documentário", ou quiserem animar um documentário histórico já decupado no Premiere, mesmo sem citar a skill. Also for English requests - animated history documentary, editorial collage motion graphics, voice-synced Remotion overlays for Premiere.

Motion Canvas — code-driven 2D explainer/animated videos in TypeScript using generator scenes (makeScene2D, yield*, refs, signals, all/chain/sequence/waitUntil), a live editor, and rendering to image sequence or MP4 (@motion-canvas/ffmpeg). Use when the user asks for Motion Canvas, or for a voiceover-synced explainer animation built outside Remotion.

Build a 3D shot for a video - live in the composition with Three.js (engines/three.js) or as a pre-rendered Blender plate (engines/blender). Covers toon, photoreal and neon looks, product and CAD models, camera moves, physics, and keeping every frame deterministic.

根据用户敲定的口播内容和手头的素材设计或改进口播科普动画。可从选题、稿子、旁白与字幕或已有视频开始，按需设计分镜、制作可编辑动画、优化排版与同步，或导出成片和发布素材。适用于知识讲解、原理演示、产品介绍与观点评论。单独转写音频、普通实拍剪辑或只做海报时不必使用。

Make a marketing video, reel, ad, post or carousel for any brand in brands/. Starts as a conversation with the person (who it's for, the idea, voice, face, look), checks the idea against what makes reels work (hook in the first second, a voice, a face, the real product, a change every second, worth sending), then storyboard, approval, production, frame-by-frame QA and delivery. Use for any "make a reel / ad / post / video for <brand>" request.

Animate a 2D cartoon character or mascot in a video - an SVG rig drawn in code (engines/svg-rig.js), a designer's Lottie file, or a Rive file. Covers part naming, idle/blink/talk/wave/hop, lip sync to a voiceover (Rhubarb mouth shapes, also on a mascot that only exists as a PNG), and the brand's mascot rules.

Rules for writing and animating short explainer videos (YouTube Shorts, Reels, TikTok and longer explainers) that people actually watch. Use when planning, scripting, storyboarding, animating or reviewing an animated explainer video, its hook, pacing, camera, characters, captions or voiceover.

Make a narrated learning-explainer video (MP4) that teaches a concept, in one soft 3D clay look with a silent mascot, with its script checked by one agent before the voice is made. Use when the user asks for a video, an explainer, an animation or a visual walkthrough of an idea, such as "make a video explaining eigenvectors", "generate an explainer on I2C", "animate how attention works". Uses what the current conversation and project already show, so run it in the conversation where the topic came up.