Video & Animation

拉片：把一条成片拆成逐镜头的分析表——每个镜头的时长、景别、类别、运镜、画面。 分工刻在骨子里：**能量的都由代码量**（切点来自 ffmpeg 场景检测，时长是切点相减， 运动量是逐帧差分的中位数），模型只判断它真正该判断的四件事（景别 / 类别 / 运镜 / 画面）， 然后每一条判断都被代码当场对账——**声称推拉摇移却实测几乎不动，门直接拦**。 看片走联系表（每镜起手帧 + 收尾帧各拼一张大图，一屏二十几个镜头，a/b 对照就是运镜）， 不是一张张翻。检测漏刀多刀用 recut 补刀并刀，自动重编号重算时长，手改边界过不了门。 产出 shots.json + Markdown 镜头表 + **单页交互式拉片报告**：内嵌播放器（播放时同步高亮镜头、 点镜头跳转）、镜头节奏带、可搜索可筛选可排序的镜头表（列表 / 卡片两种视图、首尾关键帧并排、 点图开大图）、景别类别运镜分布、出场人物、质量门、导出 JSON。单文件零依赖，离线双击能开。 14 道质量门全部由脚本确定性检查。 零依赖、零 API key，只要 node 和 ffmpeg。 Use when asked to 拉片

MiniMax multimodal model skill — use MiniMax Multi-Modal models for speech, music, video, and image. Create voice, music, video, and images with MiniMax AI: TTS (text-to-speech, voice cloning, voice design, multi-segment), music (songs, instrumentals), video (text-to-video, image-to-video, start-end frame, subject reference, templates, long-form multi-scene), image (text-to-image, image-to-image with character reference), and media processing (convert, concat, trim, extract). Use when the user m

Tanıtım/eğitim videosu üretme ve mevcut videoyu KARE DÜZEYİNDE düzeltme kiti (Python + OpenCV + ffmpeg; Android çekimi için adb). BU SKILL'İ KULLAN: Android uygulamasından ekran kaydı alıp dokunuş işaretli, açıklama panolu 1920x1080 tanıtım videosu kurgularken; zaman kodlu seslendirme metni / SRT altyazı üretirken; dikey (9:16) videoyu yatay (16:9) yaparken; videodaki bir bölgeyi başka bir çekimden taşırken (eğik baş, bozuk yüz); dönüş/hareket yönünü ters çevirirken; dizüstü ekranını tablet+parm

Process media with FFmpeg (video/audio), ImageMagick (images), RMBG (AI background removal). Use for encoding, format conversion, filters, thumbnails, batch processing, HLS/DASH streaming.

Interacting with Mediabunny

Preview a Remotion video

Export a Remotion video

Structure Remotion markup for interactivity

Build an app with Remotion

Elite frontend orchestrator — slams the full design stack together for Awwwards-tier, blow-my-mind UI work. Use when the user asks for an elite/premium/showcase/standout/hero/jaw-dropping/portfolio-grade/award-worthy frontend, a landing page that has to wow, a client demo that needs to land, a brand site, a marketing hero, a creative coding piece, an interactive showcase, a Remotion video composition, or any single-shot HTML/React artifact where the bar is craft, not just function. Composes fron

Upgrade Remotion, and related packages

Portable checklist for gating public-facing copy before release. Use when captions, posts, manifests, topics, titles, or launch copy must be checked for generic AI wording, leaked internal labels, forbidden experiment or tool names, platform topic allowlists, CTA spam, and unsupported overclaims.

Use when a dw MCP server is connected and the user wants music from MiniMax Music 3 - a song with sung lyrics, an instrumental, a score to lay under a film or a cuts piece, or the soundtrack a music video is cut to. Picks the template for the shape, states the ceiling and tag rules that bite, quotes cost, and points at MiniMax's own caption skill for the prompt.

Create, revise, review, and export Envitefy marketing videos using the local campaign engine, brand preferences, real product demonstrations, and saved campaign history. Also prepare captions and social copy for those videos.

Turn video and audio into text — full transcripts, timestamped segments, SRT and VTT subtitles, and the first-3-seconds hook — from one YouTube video, Short or live VOD, a whole YouTube channel in a single run, a creator's Instagram reels, a podcast episode, a Loom or a Twitch VOD, or any direct media file link. Captions are used where they exist and speech-to-text runs where they do not, in ~90 languages, with no third-party API key and no charge when a video has nothing to say. Use when the us

Join a Google Meet call, transcribe live captions, optionally speak in realtime, and do the followup work afterwards. Use when the user asks the agent to sit in on a meeting, take notes, summarize, respond in-call, or action items from it.

Vision-language chat: VQA, captioning, image dialogue.

Batch-research the latest 1–N videos of a YouTube channel and produce ONE cross-video comparative report (not N isolated summaries) — enumerate uploads, fetch text with captions first and local ASR as fallback, then write up themes, drift, contradictions and limits. 批量研究 YouTube 频道最近的 1 至 N 个视频：枚举投稿、逐视频取文本（字幕优先、无字幕自动 ASR 转写），再横向对比成文，产出跨视频研究报告。Trigger phrases / 触发词：研究这个YouTube频道, 频道最近的视频, 批量总结YouTube, youtube频道调研, 这个频道最近N个视频讲了什么, 频道内容分析, 竞品频道追踪, youtube 批量字幕/转录, youtube channel research, latest v

Multi-platform channel/video batch research — enumerate a creator's latest N items on YouTube, Bilibili, or local files (including Douyin creator-center exports), fetch text with captions first and local ASR as fallback, then write ONE cross-video report instead of N isolated summaries. 多平台批量研究：枚举 YouTube / B站 / 本地文件（含抖音本人账号导出）的最近 N 条内容，字幕优先、ASR 兜底取文本，产出跨视频研究报告。Trigger phrases / 触发词：批量研究这个频道, 频道最近的视频, 批量总结YouTube, B站UP主批量分析, youtube频道调研, 竞品频道追踪, 抖音账号数据复盘, 多平台内容研究, channel research, batch summari

Use when writing or editing prose people will read, including documentation, plans, messages, comments, reports, manuscripts, abstracts, captions, and research summaries.

Humanizer that writes prose (or rewrites it) so it doesn't read as AI — using an editable house-rules file plus a deep detection engine. Use whenever the user wants to de-AI any writing — a chat message, DM, email, Slack reply, social caption, or a full blog post, article, newsletter, essay, or landing-page copy. Triggers on "anti-ai", "humanize this", "make this sound human", "rewrite so it doesn't sound like ChatGPT", "less AI / less ChatGPT-y", "de-AI this", "remove AI-isms", "audit this for

Meta video ad performance audit for Reels, Stories, and in-stream placements. Evaluates hook rate (3-second view rate), hold rate, ThruPlay rate, average watch-time percentage, and the retention drop-off curve, plus sound-on/caption compliance and safe-zone spec compliance. Distinct from meta-creative's diversity/fatigue scoring — this is video-mechanics performance, not concept diversity. Use when user says video ad performance, hook rate, retention rate, watch time, ThruPlay, drop-off curve, R

Agent Media OS, the single skill for every media need in a HyperFrames project. Resolve BGM, SFX, image, icon, brand logo, voice, color grade, or LUT into a frozen local file or paste-ready block + ledger record (one verb, `resolve`); generate via TTS / music / image models when the catalog misses; produce voiceover, transcription, captions, and background removal through one shared audio engine; operate on media (cut / reframe / transform); and reuse assets across projects. Also use for vague f

Add captions or subtitles to an existing single-subject talking-head video without editing the footage. Use for plain verbatim captions, cinematic captions embedded behind the subject, VFX captions, “炸/特效/酷炫字幕,” or a named identity from the 35-style catalog. Route by visual identity, not by backend engine. The quiet `anchor` rail is the default; embed every word only when the user explicitly wants a fully cinematic treatment. The workflow runs locally end to end, including transcription and subj