Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
ryannli avatar

Immersive Reading

  • 45 repo stars
  • Updated July 1, 2026
  • ryannli/immersive-reading

Immersive Reading is a Claude Code agent skill that converts any long-form source - essay, post, transcript, note, paper, or URL - into a polished, self-contained interactive reader instead of a summary.

About

Immersive Reading is a Claude Code agent skill from Ran Li that converts any long-form source - an essay, blog post, transcript, note, paper, or URL - into a polished, self-contained interactive reader instead of a summary. The agent reads the source and builds structure (chapters, section beats, anchor quotes, preserved media) while bundled scripts scaffold, schema-validate and smoke-test a static reader template, so you get a consistently high-quality result without redesigning any UI. The output is a co-reading study space: full-text search, highlights, copyable notes, source attribution, optional bilingual mode, and light/dark themes - designed so you still read, digest and think for yourself, just in a far more engaging and navigable way. It installs as a Claude Code plugin or via npx skills add ryannli/immersive-reading, natively supports Codex, Cursor, Antigravity and other agents, ships static output ready for local use or Vercel, and is MIT licensed.

  • Not a summary - a co-reading space that keeps you doing the thinking
  • Turns any essay, transcript, paper, note or URL into a chaptered interactive reader
  • Ships full-text search, highlights, notes, source attribution, bilingual mode and light/dark themes
  • Bundled reader template + scaffold/validate scripts: the agent fills content, not UI
  • Static, self-contained output ready to run locally or deploy to Vercel; MIT licensed

Immersive Reading by the numbers

  • Data as of Aug 1, 2026 (Skillselion catalog sync)
npx skills add https://github.com/ryannli/immersive-reading --skill immersive-reading

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
repo stars45
Last updatedJuly 1, 2026
Repositoryryannli/immersive-reading

What it does

You have a dense essay, paper, transcript, or long post and want to actually digest it - turn it into a beautiful interactive reader with chapters, search, highlights and notes, not a lossy TL;DR.

Who is it for?

Readers, researchers and builders who want to deeply digest a specific long essay, paper, transcript or article - and keep a reusable, navigable study space rather than a throwaway TL;DR.

Skip if: Getting a quick one-paragraph gist, bulk-summarizing a feed, or any task where you do not actually want to read the source - this keeps you in the reading, it does not replace it.

When should I use this skill?

When you want to turn a blog post, essay, transcript, note, paper, or URL into a chaptered static reader with search, highlights, notes, bilingual support and themes.

What you get

You get a beautiful chaptered reading space with search, anchor quotes, highlights, notes and source attribution, so you actively read and retain long-form ideas in a much more engaging way.

  • a self-contained static interactive reader site (chapters, search, highlights, notes, attribution)
  • validated article-data.js matching the reader template schema

By the numbers

  • MIT licensed
  • 26 GitHub stars (launched 2026-06-30)
  • installs as a Claude Code plugin or via npx skills add

Files

SKILL.mdMarkdownGitHub ↗

Immersive Reading

Overview

Create a self-contained static reading site from new source material by reusing the bundled reader template. Do not redesign the UI from scratch unless the user explicitly asks; the template already encodes the polished interaction details.

The skill separates judgment from deterministic work:

  • Use the model for source structure, chapter/section titles, quotes, summaries, translation, and editorial judgment.
  • Use bundled scripts for scaffolding, schema validation, and smoke tests.

Workflow

1. Intake Ask only for missing essentials:

  • source content, source URL, or source file
  • whether to include bilingual mode and target language, if not already clear

Do not ask for an output folder by default. Choose a sensible local folder name from the source title or article slug, then tell the user where the site was created. Only ask for a destination when the user has special placement or deployment requirements.

2. Source Handling Treat the project as a local reading/study space by default. Do not block the workflow with ownership questions. Preserve source title, author, and original link when they are known. If the original source contains meaningful media (images, diagrams, screenshots, embeds, audio, or video), preserve it as part of the learning site. Prefer downloading stable media into the generated site's assets/source/ folder; otherwise keep the original media URL. Add the media to the relevant section or paragraph data with alt text, captions, and source credit when available, then verify it renders in preview.

3. Content Conversion Read references/content-model.md and produce one article-data.js file matching the template schema. If the source lacks structure, create 5-14 chapters and 1-4 sections per chapter. Each section needs:

  • take: clear section title
  • quote: short anchor quote or excerpt
  • paragraphs: source text, summary text, or translated/source-aligned text
  • optional footnotes

4. Scaffold Run:

   node skills/immersive-reading/scripts/scaffold-reader.mjs --article-data /path/to/article-data.js --out /path/to/output

5. Validate Run the generated-site checks:

   node skills/immersive-reading/scripts/validate-article-data.mjs /path/to/output/src/articles/<article-id>/data.js
   node skills/immersive-reading/scripts/smoke-test-reader.mjs /path/to/output

If you changed this skill's template, references, or scripts, also run:

   node skills/immersive-reading/scripts/check-skill.mjs

6. Preview Serve the generated folder with the bundled local server, inspect desktop and mobile, and fix data/layout problems before delivery. Use this server for preview because it supports byte-range video seeking required by the scroll-scrub opening:

   node skills/immersive-reading/scripts/serve-reader.mjs /path/to/output --port 8791

7. Deploy Read references/deployment.md only when the user asks to publish.

Template Boundary

Use assets/reader-template/ as the source of truth for the generated site. The template intentionally excludes production analytics, dashboards, Upstash/Vercel secrets, and project-specific branding.

Preserve these features unless the user asks otherwise:

  • chapter/section reading flow
  • search
  • highlights
  • notes and copyable notes
  • optional bilingual display
  • light/dark mode
  • mobile limited-functionality notice
  • source attribution and original link
  • 3D particle background and scroll-driven transitions

Read references/design-contract.md before changing template layout, animation, or visual hierarchy.

Related skills

How it compares

Use instead of pasting an article into a chatbot for a TL;DR: it preserves the full source as an interactive reader with structure and study tools, rather than compressing it into a lossy summary.

FAQ

What is the Immersive Reading skill?

A Claude Code agent skill that turns any long-form source (essay, post, transcript, note, paper, or URL) into a self-contained interactive static reader with chapters, search, highlights, notes, source attribution, optional bilingual mode, and light/dark themes.

How do I install it?

In Claude Code run /plugin marketplace add ryannli/immersive-reading then /plugin install immersive-reading@immersive-reading, or install it as an open skill with npx skills add ryannli/immersive-reading. It also works with Codex, Cursor and other agents. MIT licensed.

Is it just an AI summarizer?

No - it is explicitly not a summary. It builds a co-reading space that keeps the full source and adds structure and study tools (chapters, quotes, search, highlights, notes) so you do the reading and thinking yourself, in a more engaging format.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.