Library OS turns the books you read into deep-dives on your own SEO-indexed website — with quotes, chapter summaries, JSON-LD schema, and the connections between ideas. Free open-source foundation. Paid leverage on top.
Live on frankx.ai/library · this is the canonical reference implementation
The Problem
You highlight in Kindle. Sync to Readwise. Tag in Notion. Mind-map in Obsidian. And then — silence. The insights compound nowhere. You can't link to them. They don't rank. They don't bring you readers. They die when you cancel the subscription.
Goodreads gets the traffic. BookTok gets the attention. You get the receipts in someone else's database.
Library OS is the opposite move. Every book you read becomes a deep-dive on your own domain, with your own JSON-LD schema, ranking for your own keywords. Permanent. Public. Yours.
How it works
/library-addAdd any book in 30 seconds. TL;DR, 5 insights, FAQ, cover. Idempotent — run it on a book that already exists, it just updates.
/library-deepenPulls quotes (with chapter + context) and chapter-by-chapter summaries. Anti-hallucination guardrails: paraphrases go in `context`, never in `text`.
/library-researchAdds curated "Continue Reading" recommendations (with the connection stated) and "Go Deeper" video links. Never invents URLs.
Cross-AI portable. The commands ship as plain-text prompts. They work in Claude Code, ChatGPT, Codex, Gemini CLI, Cursor — or pasted by hand into any chat interface.
No lock-in. The data is plain TypeScript (data/book-reviews.ts) following a public schema. Export, migrate, fork — your library is yours.
Choose your tier
The repo is MIT — clone it and ship today. Each paid tier saves you a different amount of time. Pick the one that matches the time you don't have.
The foundation
Clone the repo, run the commands, publish your library. MIT-licensed. No strings.
Fork-and-go
A pre-configured fork seeded with 10 starter book entries and 1 deep-dive — so your library looks credible on day one.
The full system
Everything in Starter, plus 50 fully-deepened book entries — you skip 50+ hours of curation and ship an instant authority site.
I build it on your domain
I build a 50-book deepened library on your domain, voice-matched to your existing content. You wake up with an authority asset that ranks.
Multi-author libraries
For publishers, universities, and content companies running multi-author book corpora at scale.
Why open source
The Library OS repo will always be MIT. You can clone it, fork it, white-label it, sell your own service on top of it. We'd be honored.
The paid tiers exist because some people want the asset without the curation work. €197 for 50 deepened books is roughly €4 per book — versus 1+ hour per book of doing it yourself. €997 for done-for-you is the trade for thought leaders whose hour is worth more than that.
Same pattern as Tailwind, Vercel, Plausible, ConvertKit. Give the foundation. Sell the leverage. Both sides win.
Questions
Notion notes die when you cancel the subscription, change tools, or just stop opening the app. Library OS publishes them as your own SEO-indexed website with JSON-LD schema. Each book becomes a long-tail SEO asset that ranks, attracts readers, and demonstrates intellectual depth — for years.
Those are platforms — you rent attention inside someone else's app. Library OS is your own domain, your own design, your own JSON-LD, your own SEO. You own the asset. Goodreads gets the traffic; Library OS gets you the traffic.
You need to be comfortable with one terminal command (clone) and one click (Vercel deploy). If you can publish a Next.js site, you can run Library OS. If you can't — that's exactly what the Starter Kit and Done-For-You tiers exist for.
The repo is the foundation; the paid tiers are leverage on top. Same model as Tailwind, Vercel, and Plausible — give the tool, sell the time-savings. The OSS users become advocates; the paid users save 50+ hours of manual curation. Both win.
Yes — frankx.ai/library is the canonical reference. 22 books, 22 fully deepened, 277+ quotes, 248+ chapter summaries. Every page on this site uses the same schema and template you get with the OSS repo.
The Library OS workflow has anti-hallucination guardrails baked in: never fabricate verbatim quotes — paraphrases go in `context`, not `text`. Never invent video URLs. Every recommendation in "Continue Reading" must state the connection explicitly. The system enforces this through the prompt design, not just goodwill.
Starter Kit and Curator's OS open in waves. Drop your email — first wave gets founder pricing and lifetime updates.
No spam, ever. Unsubscribe anytime.
No spam. One email when the kits open. Unsubscribe anytime.