Communitygithub.com

interactive-to-video

把动页(可交互的网页作品、互动文章、explorable explanation)做成完整的讲解视频:为这件动页写一份录制脚本,无头浏览器照着脚本像读者一样把页面走一遍并录屏,配上克隆音色的旁白或页面自带的声音,出 1080p mp4 和字幕。用户说「把这个动页 / 互动网页做成视频」「动页转视频」「出个视频版」时用。

interactive-to-video 是什么?

interactive-to-video is a Claude Code agent skill that 把动页(可交互的网页作品、互动文章、explorable explanation)做成完整的讲解视频:为这件动页写一份录制脚本,无头浏览器照着脚本像读者一样把页面走一遍并录屏,配上克隆音色的旁白或页面自带的声音,出 1080p mp4 和字幕。用户说「把这个动页 / 互动网页做成视频」「动页转视频」「出个视频版」时用。.

兼容平台✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/micahfang1/interactive-to-video/tree/HEAD

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Interactive Page to Video

This directory is the tool itself. A headless browser walks through the interactive page like a reader (scrolling, clicking, dragging – cursor is visible), recording the screen along the way, and produces a 1080p MP4 with voiceover.

Recording and compositing are automatic; you must write the script for each interactive page: understand the page, decide what to cover, what to cut, and what actions to perform on each step. This is the main effort.

  • lib/record.mjs – Recorder: opens the page, injects cursor and subtitles, captures frames into a video, records the page's own audio.
  • lib/make.mjs – All-in-one for explainer mode: synthesize voiceover → drive interactions synchronized with voiceover duration → compose final video and subtitles.
  • lib/voice.mjs + lib/tts.py – Voice cloning (IndexTTS-2).
  • lib/subtitles.mjs – Subtitle splitting, duration estimation, silence detection (has tests, run npm test).
  • examples/demo/ – A ready-to-run mini example (one small interactive page + its script); run it first to confirm the environment.
  • examples/ – Two more real-world examples of each type; they won't run as-is (they depend on their respective interactive page files), but show how to structure the code.

See README for environment setup. Intermediate files are written to work/, final output to out/, so run this from the directory where you want the video to appear.

Classification First

A Explainer-TypeB Self-Voiced-Type
Interactive page featuresNo voiceover, relies on reader reading and interactingHas its own audio, sound effects, subtitles, auto-plays on scroll
Video's audioWrite a separate voiceover, clone the voiceRecord the page's own audio; only add voiceover for text without audio
Exampleexamples/explainer-involution.mjsexamples/selfvoiced-jensen.mjs
Commandnode lib/make.mjs <script>.mjsnode <script>.mjs

To create a new episode: copy the relevant example and modify. The comments in the examples explain what each section does.

Principles

  • Make a complete explainer video, not a teaser.
  • Keep the pace tight: minimize silent parts. After rendering, check the list of

相关技能