ryannli/immersive-reading

Immersive Reading is a tool that transforms long, dense written content into an interactive reading experience. You give it a link, essay, transcript, note, or paper, and it produces a polished learning space with chapters, source attribution, search, highlights, notes, and optional bilingual support. The project describes itself as "not a summary" but rather a "co-reading space," meaning it preserves the original material while adding structure and tools to help you work through it. The workflow has three stages. First, you bring source material in the form of raw text, a URL, or a local file. Second, an AI agent shapes that material into a structured reading experience with chapters, section breaks, anchor quotes, and media. Third, you get an interactive learning space that includes search, highlighting, notes, and light or dark mode. The output is a static website you can run locally or deploy on a hosting service like Vercel. The same scaffolding can be reused across multiple pieces of content. Installation depends on which AI coding assistant you use. The primary path is Claude Code, where you install it as a plugin through a series of commands. It also supports Codex, Cursor, and an "Antigravity CLI," plus any other agent that works with the SKILL.md format. Most installations use a command line tool called npx skills add that pulls the skill folder directly from GitHub, so the project does not need to be installed as an npm package. To use it, you tell your AI agent to apply the immersive reading skill to an article or link. You can request bilingual mode by asking for a specific language, such as Spanish. You can also paste text directly or point to a local file. The agent decides where to create the site and tells you when it is ready. The README includes a demo based on Paul Graham's essay "How to Do Great Work," showing the before and after. The project is MIT licensed.