PitterLopes

AI Reels Editor Skill — Data skill for Claude Code

Data community

Claude Code skill: turn a talking-head video into an edited Reel/TikTok/Short — silence cuts, word-timed kinetic captions, words behind the speaker, cards, animated data, SFX.

How to install AI Reels Editor Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open PitterLopes/ai-reels-editor-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What AI Reels Editor Skill does

Claude Code skill: turn a talking-head video into an edited Reel/TikTok/Short — silence cuts, word-timed kinetic captions, words behind the speaker, cards, animated data, SFX. HyperFrames + ffmpeg + Whisper. EN/PT-BR.

Alternatives in Data

HTML Anything — ✨ The agentic HTML editor — your local AI agent writes the HTML, you ship it 8.5k ★
Pipeline Feature — Add a feature to the FFmpeg transcode pipeline following Tunarr's pipeline builder pattern 2.5k ★
Motion Skills — 50 open-source skills that teach your AI coding agent to make motion graphics, animation & video — kinetic typ 477 ★

README

AI Reels Editor — a Claude Code skill that edits talking-head videos into motion-graphics Reels

**English** · [Português](README.pt-BR.md)

`reels-editor` is a [Claude Code](https://claude.com/claude-code) skill (also usable with the Claude Agent SDK) that turns a raw talking-head take into an **edited vertical video for Instagram Reels, TikTok and YouTube Shorts**: silence cuts, kinetic captions, words that appear *behind* the speaker, hard reframes (no in-and-out zoom), presentation cards, animated numbers and synthesized sound effects — rendered with [HyperFrames](https://github.com/heygen-com/hyperframes) (HTML + GSAP, open source).

The key idea: **the AI does not write animation code per video.** It writes a short *plan* (what to show, when, in which slot), the plan is validated in code, and a fixed, measured **style** turns it into a composition. Twenty styles are specified in [`references/styles.md`](references/styles.md); three are implemented today (`impact`, `card`, `data`).

Built for the [Scriptroom](https://scriptroom.online) app ("AI script writer + teleprompter"). This repository is the open, chat-testable version of that editing concept.

What you get

Cuts: pauses above 0.6 s removed, 0.2 s of air kept, frame-accurate (CFR 30 fps).
Transcript with word timings (Whisper, local) → captions never say a word that was not spoken.
Speaker cutout (optional, hyperframes remove-background) so big words can sit between the speaker and the background.
Head tracking from the alpha mask: words behind the head land where the head is, per second.
Safe zone for Reels (top 220 px, bottom 1470 px, right column) respected by every element.
Hard reframes held (1.0 → 1.18 → 1.0): never a zoom that goes in and comes back.
Presentation cards (the speaker becomes a framed card, bullet points appear with the voice), stat and compare cards.
Synthesized SFX (whoosh, pop, click, thump, sparkle) on their own