🎬 Claude Remotion Skill — Make AI Video Editing & Motion Graphics with Claude

An open-source Claude agent skill that turns Claude Code, Claude Desktop, and Claude AI into a real motion-graphics editor. It teaches Claude to create and edit professional AI videos with Remotion — springs, staggered choreography, color grades, film grain, Ken Burns, word-synced captions, and sound design — from a single prompt. Text to video, the way an actual editor would do it.

If you've ever asked an AI to "make a video" and gotten flat fades and linear motion, this is the fix. Remotion is never the bottleneck — motion-design craft is. This skill encodes that craft as hard rules plus a mandatory render → inspect frames → fix → re-render loop, so Claude verifies its own output instead of shipping blind.

🎥 Demo — made with this skill

A motion-graphics clip produced by Claude Code using this skill. ▶️ Watch/download the full-quality MP4 (with audio)

🤖 What is this? (TL;DR)

Claude Remotion Skill is a free, open-source agent skill for Anthropic's Claude. You install it once, then ask Claude — in plain English — to make or fix a video, and Claude renders a finished, professionally animated clip using Remotion (React + TypeScript for video). It handles motion design, B-roll, captions, transitions, and sound design automatically.

❓ FAQ

Can Claude edit videos? Yes. Out of the box Claude can write video code; with this skill it produces professional motion graphics — proper easing, staggered timing, color grading, grain, captions — and verifies the render by extracting and inspecting frames before delivering.

Do I need After Effects or Premiere? No. Everything runs in code via Remotion (React + TypeScript). No timeline software, no manual keyframing. You describe the video; Claude builds and renders it.

Which Claude models work? Claude Code and Claude Desktop on Sonnet or Opus. The skill is model-agnostic — any Claude that supports agent skills can use it.

Is it free? Yes. MIT licensed, fully open source. You only need Remotion (free, open source) and a Claude subscription/API access.

What can it make? Logo animations, brand intros, vertical Reels/Shorts/TikToks, faceless content videos, product explainers, animated captions/subtitles, lower thirds, kinetic typography, and B-roll-driven edits.

How is this different from text-to-video models (Sora, Runway, etc.)? Those generate pixels and hallucinate. This produces deterministic, editable, brand-controlled motion graphics from real code — exact text, exact colors, exact timing, re-renderable forever.

🆚 Generic AI video vs. with this skill

Without this skill	With Claude Remotion Skill
Linear interpolation everywhere	Springs + bezier curves only (easeOutExpo, presets)
Lonely opacity fades	3-property entrances (fade + rise + scale)
Everything enters at once	Staggered choreography (3–6 frame offsets)
Flat solid backgrounds	Drifting gradient mesh + grade + grain + vignette stack
Static pasted images	Ken Burns on every still, OffthreadVideo for footage
Ships unverified renders	ffmpeg frame extraction + visual inspection, every time

📦 What's inside