ajaygunalan

Chapter To Synced Lecture — Development skill for Claude Code

Development community

Turn a book chapter (PDF) into a lecture given by its author: HTML slides driven by narration, questions the audio stops on, word-synced captions.

How to install Chapter To Synced Lecture

This entry records only its repository, not the path inside it, so there is no exact command to give. Open ajaygunalan/chapter-to-synced-lecture and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Chapter To Synced Lecture does

Turn a book chapter (PDF) into a lecture given by its author: HTML slides driven by narration, questions the audio stops on, word-synced captions. Local GPU voice (Kokoro), free and unlimited. A Claude Code skill.

Alternatives in Development

Ars Plan — ARS academic-paper plan mode — Socratic chapter-by-chapter planning 4.5k ★
Claude Reviews Claude — Claude reads its own source code — 17-chapter architectural deep-dive into Claude Code v2.1.88. EN/ZH bilingua 1.3k ★
Chapter Outline — 生成章纲 skill 669 ★

README

chapter-to-synced-lecture

A [Claude Code](https://claude.com/claude-code) skill that turns a book chapter (PDF) into a **video lecture**: narration plus slides that animate along with it, given in the voice of the book's own author.

It comes out as one self-contained web page you open in a browser — not an MP4, but it plays like one, and you can pause, scrub, jump to any slide, or read the transcript.

**No API key needed.** The voice runs on your own machine with [Kokoro](https://huggingface.co/hexgrad/Kokoro-82M): free, unlimited, about a minute of GPU time for an hour of narration. ElevenLabs is optional, if you want a paid voice for a final take.

What you get

Slides that animate with the voice. When the narration says "A to B, for five," that edge lights up at that word.
Questions it stops on. The audio pauses on a question and waits. Press play when you have an answer.
Captions in sync, word by word, plus a full transcript you can click to jump anywhere.
Slides that show real runs. If the chapter has an algorithm, the page contains a working copy of it and runs it to make the slides. So the picture can never disagree with the algorithm, and the narration is written from what the slides actually produced.

Then you revise it with the same command

Point it at the same PDF and it finds the lecture already there, reads it, and gets ready to teach. Go through a tab, say what lost you, and it explains that bit in the chat first — then edits the script or the slide once you're happy, and re-records just that part. A few seconds, because each tab is its own audio file. Nothing else is touched.

How it works

Five steps, from PDF to a lecture open in your browser:


Read	pull out the text and figures, and render any page the text layer mangles
Plan	work out what each part is for: the problem it opens on, the example that carries it
Slides	build the deck
Script	writ