convez.ai
SkillEscrita & ConteúdoIntermediário
frankxai

Library OS

frankxai/library-os

Converta leituras em biblioteca permanente e citável com IA

Ver no GitHub
InicialPluginScore · de 0 a 100

Sistema inteligente de biblioteca pessoal que captura, extrai e publica análises profundas de cada livro que você lê em seu próprio website. Converte highlights, notas e aprendizados em páginas permanentes, ricamente estruturadas e otimizadas para SEO, transformando leitura dispersa em conhecimento composto.

Instalação

$npx convez-ai frankxai/library-os

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
2
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 3 semanas
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Criar biblioteca pessoal estruturada a partir de Kindle highlights e anotações
  • Gerar páginas de deep-dive automáticas com TOC, quotes e FAQs
  • Publicar análises de livros com schema JSON-LD para SEO e APIs

Veredito

Solução robusta e opinionada para quem quer transformar leitura em ativo digital permanente. Bem documentada, deployment direto em Vercel, integra nativamente com Claude via plugin — recomendado para produtores de conteúdo e pesquisadores que precisam compor conhecimento acumulado.

README

A tradução para o português deste README ainda está na fila.

Library OS

The open-source Library Intelligence System. Capture, extract, enrich, and publish every book you read into a permanent deep-dive on your own website.

Deploy with Vercel License: MIT Claude Code Plugin

Live reference: 22 books, 277 curated quotes, dynamic OG images, JSON quote API at frankx.ai/library · The story: frankx.ai/library/approach

Install (5 paths, pick one)

# Bootable Next.js site
git clone https://github.com/frankxai/library-os.git my-library && cd my-library && npm install && npm run dev

# Or — Claude Code plugin only (commands + skill + subagent)
rsync -av <(curl -sL https://github.com/frankxai/library-os/archive/main.tar.gz | tar xz)/library-os-main/.claude/ .claude/

# Or — via ACOS
npx @frankx/agentic-creator-os init  # library-os included from v11+

Full install paths in INSTALL.md — bootable site, drop-in to existing Next.js, plugin-only, cross-AI (ChatGPT/Cursor/Codex/Gemini), via ACOS.


Why

Readers forget up to ninety percent of what they read within a week. Kindle highlights die in Kindle. Notion docs scatter. Goodreads knows what you read but not what you took from it.

Every book you've read is capital. Without a system to compound it, that capital depreciates to zero the moment the book goes back on the shelf.

Library OS is the system that makes it compound. Every book becomes a permanent, intelligent, schema-rich page on your own website — TOC'd, quoted, chapter-distilled, FAQ-answered, cross-linked, SEO-ready, LLM-citable. In six months you own a library that compounds for decades.

What you get

  • Data schema — one TypeScript interface for everything a book can be
  • Next.js App Router template — /library index + /library/{slug} deep-dive + /library/approach showcase
  • Three slash commands — /library-add, /library-deepen, /library-research
  • One skill — library-os (documents the full workflow)
  • One subagent — book-distiller (the extraction specialist)
  • Full JSON-LD schema — BreadcrumbList + Article + Review + FAQPage + Quotation on every book page
  • Two example deep-dives — Profit First + The Fabric of Reality (ship real books in under an hour)

Quick start (5 minutes to live)

# 1. Clone
git clone https://github.com/frankxai/library-os.git my-library
cd my-library

# 2. Install
npm install

# 3. Run
npm run dev
# open http://localhost:3000/library

Edit data/book-reviews.ts to add your first book. Redeploy via git push — Vercel picks it up.

Alternative: cherry-pick into your existing Next.js project

If you already have a Next.js App Router site, copy these files into it:

app/books/types.ts
app/books/lib/books-registry.ts       ← customize for your own books
app/library/page.tsx
app/library/[slug]/page.tsx
app/library/approach/page.tsx         ← optional showcase page
data/book-reviews.ts                  ← your data
public/images/library/                ← book cover JPEGs
.claude/commands/library-*.md         ← the three slash commands
.claude/skills/library-os/SKILL.md    ← the workflow skill
.claude/agents/book-distiller.md      ← the extraction subagent

Requires: Next.js 14+ App Router, Tailwind CSS v3+, TypeScript. No other dependencies.

The four-stage workflow

CAPTURE ──────► EXTRACT ──────► ENRICH ──────► PUBLISH
   │                │                │              │
   ▼                ▼                ▼              ▼
Photo, notes,   quotes[]      continueReading[]   /library/{slug}
Kindle export,  chapters[]    videos[]            (live, SEO+AEO)
voice memo      tldr, faq     + schema

1 — CAPTURE. A photo of handwritten margin notes, a Kindle highlights export, a voice memo, a book title in a text file. You refuse to let a thought leave the system.

2 — EXTRACT. Run /library-add "Book Title" to create the baseline entry with TL;DR, 5 key insights, Best-For audience, starter FAQ, and a book cover pulled from OpenLibrary.

3 — ENRICH. Run /library-deepen {slug} to populate 10–20 curated quotes and every chapter. Then /library-research {slug} to add Continue-Reading and Videos.

4 — PUBLISH. git commit && git push. Vercel redeploys. Your book is now a permanent URL with the full deep-dive hub + rich structured data.

The data schema

app/books/types.ts — the spine of the system.

interface BookReview {
  // Identity
  slug, title, author, coverImage, rating, reviewDate, categories, readingTime

  // Core content (required)
  keyInsights: string[]   // exactly 5
  bestFor: string[]       // 3–4

  // AEO layer (optional, strongly recommended)
  tldr?: string
  faq?: Array<{ q, a }>
  publicationYear?: number

  // Deep-dive layer (optional — progressive depth)
  quotes?: BookQuote[]                  // text, chapter, context
  chapters?: BookChapterSummary[]       // number, title, keyIdea, summary
  continueReading?: RelatedReadingItem[]
  videos?: BookVideo[]

  // Wiring
  amazonUrl?: string
  relatedBook?: string    // links to your own books via booksRegistry
  hasCover?: boolean      // gates next/image render
}

Books without optional fields render as concise reviews. Books with all fields render as full knowledge hubs. One template, progressive depth.

The three slash commands

CommandPurposeWrites to
/library-addCreate a new entry from just a titledata/book-reviews.ts, public/images/library/
/library-deepenAdd quotes + chapters to existing entrydata/book-reviews.ts
/library-researchAdd continueReading + videosdata/book-reviews.ts

Run in order. Each is idempotent. Full documentation in .claude/commands/.

Cross-AI compatibility

ToolNative supportHow to use
Claude Code✅Slash commands + skill + subagent all work out of the box
ChatGPT / Claude.ai (web)—Paste the prompt from .claude/commands/library-*.md, paste book details, get structured TypeScript back
Cursor / Codex / Gemini CLI—Read the command files as plain-text instructions, execute the steps
Pen & paper—The schema and template work without any AI. A hand-curated library is always valid.

See docs/cross-ai-guide.md for paste-ready prompts.

Deployment targets

Library OS ships as a Next.js App Router app, but the spine (the BookReview TypeScript type + the JSON-LD schema emission) is framework-neutral.

StackPort effortNotes
Next.js App RouterZeroDirect. This repo.
Astro / StarlightSmallPort app/library/ to src/pages/library/. Schema unchanged.
SvelteKitSmallroutes/library/+page.svelte and +page.svelte in [slug].
Hugo / Jekyll / 11tyMediumEach BookReview becomes a markdown file with frontmatter.
Plain HTMLMediumA small build script renders static pages. Schema works standalone.

Documentation

Contributing

Issues and PRs welcome. The spirit of Library OS is curation over automation — AI helps you move faster, but the final judgment on what deserves to be in your library is yours.

License

MIT — see LICENSE.

Credits

Built and maintained by Frank. Reference library at frankx.ai/library.

Inspired by the intuition that the best tools for thought are the ones you own and the ones that outlive their vendors.

Ver o original no GitHub

Bom para

Relacionados