Video-Editing

Research/2026-05-03-claude-video-skills-landscape.md

Claude Code Video & Animation Skills — 2026 Landscape

TL;DR

Claude Code has a mature ecosystem for video and animation work. Dominant pattern: install a Skill (folder of instructions + reference docs) into ~/.claude/skills/, then prompt in natural language. Output ranges from React-rendered MP4s to Sora 2 / Veo 3.1 cinematic clips to mathematical animations to logo motion. Breakthrough was the official Remotion skill (Jan 2026, 6M+ launch views, 25k+ installs week one).

1. The Big One: Remotion Skill Family

Remotion = "React for video." You write React/TypeScript components describing each frame; it renders to MP4. Claude writes the components.

Official skill install:

npx skills add remotion-dev/skills
# or during scaffold:
bun create video

The official package contains 28 modular rules covering animations (interpolate(), spring(), useCurrentFrame()), compositions (<Sequence>, <AbsoluteFill>), captions, charts, lottie integration.

Notable community Remotion skills:

Skill	What it adds
wshuyi/remotion-video-skill	MiniMax / Edge TTS for voiceovers, scene-based timing
jhartquist/claude-remotion-kickstart	Scaffolded starter project
affaan-m/everything-claude-code — `remotion-video-creation`	`lottie.md` rules, asset pipeline
digitalsamba/claude-code-video-toolkit	Brand profiles, transitions, reusable component library

What people are actually building (from sabrina.dev and louisedesadeleer.substack.com):

30-sec Kurzgesagt-style explainers (1080×1920 vertical, SVG diagrams + spring animations)
25-sec product demos — Claude scrapes URL, pulls branding/screenshots, builds simulated cursor-driven demo + real-image showcase
20-sec Google Reviews testimonial reels — real star ratings, animated 5-star reveals, carousels
Avatar/talking-head overlays — input 9:16 video, Claude transcribes, layers animated titles + keyword badges
15-sec data viz dashboards from CSV — KPI cards, animated bar/donut/line charts
Pinterest/YouTube motion graphic recreations — bar charts, Venn diagrams, kinetic title cards, growth graphs

Workflow tip: Claude can't ingest video directly. Use FFmpeg to extract frames, hand them to Claude as images, and it reverse-engineers the design into a Remotion composition.

2. AI Video Generation (Sora / Veo / Kling) Skills

Wrappers that put hosted video models behind a Claude skill so you stay in CLI:

Skill / MCP	Models	Notes
Higgsfield MCP (official, 2026-04-30)	Sora 2, Veo 3.1, Kling 3.0, 30+ models	One hosted endpoint, polling loop
Kie Video Generator	Veo, Sora, Kling	Unified interface
Evolink Media	60+ models	Sora, Veo 3, Kling
SamurAIGPT/Generative-Media-Skills	Midjourney v7, Flux Kontext, Seedance 2, Kling 3, Veo3, Suno, MMAudio	19 MCP tools, modular

SamurAI bundle sub-skills: - Cinema Director — structured cinematography prompting - Seedance 2 — text-to-video / image-to-video / video extension with native audio sync - AI Clipping — long video → ranked vertical shorts with face-tracked auto-crop - YouTube Shorts — platform-aware preset for Shorts/TikTok/Reels

3. Mathematical Animation (Manim / 3Blue1Brown style)

Multiple skills wrap Grant Sanderson's Manim engine:

Skill	Approach
Yusuke710/manim-skill	Plugin marketplace install. Autonomous Plan → Code → Render → Iterate loop
adithya-s-k/manim_skill	Agent skills for 3Blue1Brown-style
manim-claude (PyPI)	24 reference guides + safe-wrapper utilities + production templates
Math-To-Manim	Multi-pipeline (Claude + Gemini + Kimi), 55+ examples
Manim MCP Server	Server-side scene rendering as MCP service

Install Yusuke's:

/plugin marketplace add Yusuke710/manim-skill
/plugin install manim-skill/manim-skill

4. Animated Websites (3D, Scroll, Interactive)

Where viral landing pages come from. Two repos dominate:

freshtechbro/claudedesignskills — 22 skills

Core 3D: threejs-webgl, gsap-scrolltrigger, react-three-fiber, motion-framer, babylonjs-engine Extended 3D/Scroll: aframe-webxr, vanta (lightweight effects), playcanvas, pixijs-2d, locomotive-scroll, barba-js Animation libs: react-spring-physics, magic-ui, AOS, anime.js, lottie 3D authoring: blender-web-pipeline, spline-interactive, rive-interactive, substance-3d-texturing

greensock/gsap-skills — Official, 8 skills

gsap-core, gsap-timeline, gsap-scrolltrigger, gsap-plugins (Flip, Draggable, SplitText, MorphSVG, ScrambleText, GSDevTools), gsap-utils, gsap-react, gsap-performance, gsap-frameworks (Vue, Svelte).

npx skills add https://github.com/greensock/gsap-skills

Viral data: Huashu Design collected 662 stars in two days. Open-source Claude Design clones collectively passed 5,000 stars. Reports of full animated landing pages built in an afternoon for under $10 in tokens.

5. Logo / SVG / Lottie Animation

Skill	Output
talknerdytome-labs/wiggle-claude-skill	Static logo (PNG/SVG/JPG) → animated GIF/MP4 via Lottie. Patterns: fade, bounce, pulse, rotate, wiggle, waveforms
Lottie Animator	Static SVG → production Lottie JSON
Lottie & Rive Animation Expert	Lightweight interactive web/mobile animations
adityakotha03/LogoAnimation (Animify)	SVG → animated SVG
LottieFiles MCP	Claude builds animations directly inside Lottie Creator

6. Typography & Animated Text

Anthropic's official frontend-design skill (default in Claude Code) — pushes distinctive fonts (avoids Arial/Inter), recommends Motion library for React, CSS for HTML, staggered reveal patterns
sliday/google-fonts-skill — Google Fonts typography system generator
Typography Designer (mcpmarket) — design system tokens
FFmpeg Karaoke / Animated Text skill — physics-based motion (spring bounces, elastic overshoots, organic wobbles) for burned-in subtitles

7. Video Editing (FFmpeg + Real Footage)

For working with existing video files (not generation):

Skill	Capabilities
browser-use/video-use	Filler-word removal, auto color grading, burned subtitles. Talking heads, tutorials, vlogs, corporate explainers
FFmpeg Video Editor (claudeskills.info)	Cut/merge/transcode/filter, frame-accurate splits, batch
Video Processing & Editing (mcpmarket)	Subtitle filters (font/size/position), SRT/ASS/VTT, soft + hard subs
YouTube Clipper Skill (op7418)	Download → semantic chapters → segment clip → translated subs
Playwright MCP with `--save-video=1920x1080`	Browser-driven product walkthroughs/screen recordings

8. Curated Awesome Lists (discovery)

travisvn/awesome-claude-skills + ComposioHQ/awesome-claude-skills — primary directories
anthropics/skills (official) — algorithmic-art (p5.js + flow fields), canvas-design (PNG/PDF), slack-gif-creator, frontend-design
awesomeclaude.ai — visual directory
awesomeskills.dev — multi-agent (Code, Codex, Cursor)
claudemarketplaces.com — 120k devs/month
claudeskills.info — 140+ free skills
alirezarezvani/claude-skills — 232+ skills across 8 agent platforms

9. Notable Adjacent Skills

frontend-slides (zarazhangrui/frontend-slides) — animation-rich HTML decks, also converts PowerPoint
claude-d3js-skill (chrisvoncsefalvay) — D3 data visualizations
claude-youtube-skill (zerowing113) — playlists → reference articles with embedded player + clickable timestamps
Hyperframes integration — animated text + motion graphics + subtitles on existing video

10. Recommended Starting Stack

For the Video_Editing working directory specifically:

Remotion official skill (npx skills add remotion-dev/skills) — core engine
GSAP official skills (npx skills add https://github.com/greensock/gsap-skills) — for any web/landing-page motion
FFmpeg skill — frame extraction + post-processing
One AI video model bridge (Higgsfield MCP for Sora 2/Veo 3.1/Kling, or SamurAIGPT's Generative-Media-Skills bundle)
Manim skill (Yusuke710's) only if doing math/explainer content

Common pattern across successful creators: start narrow, iterate per-component, keep prompts grounded in a specific design ref. Maximalist prompts ("make it amazing") produce mush; tight prompts with one inspiration image produce ship-quality work in minutes.