Book Lover

Your AI librarian from Claude or Cursor — track reading lists, get recommendations, search 8 sources, download and convert books with 35+ tools.

An MCP server for managing your personal book library. Track what you read, write reviews, get recommendations, search across multiple sources, and convert between formats — all from your AI assistant.

Quick Start

# No API key needed — just run
claude mcp add book-lover -- uvx mcp-book-lover
# Ask Claude: "Find books by Brandon Sanderson" or "What should I read next?"

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "book-lover": {
      "command": "uvx",
      "args": ["mcp-book-lover"]
    }
  }
}

Features

• Book tracking — manage your library with statuses: want to read → downloaded → reading → finished
• Series support — track progress through book series with ordering and series-level reviews
• Library search — find books in your library by title, author, series, or description
• Reviews & quotes — rate books and series (1–5), write reviews, save favourite quotes
• Recommendations — personalized suggestions based on your ratings and series reviews; save recommendations for later
• Multi-source search — search online across 8 sources (Google Books, Open Library, Author.Today, Knigogo, Searchfloor, LibGen, Flibusta, Project Gutenberg)
• Series/cycle search — bl_search_series groups numbered parts of a cycle (e.g. "Брутфорс 1..6") into a single result
• Download — download books directly from searchfloor.org to ~/Books; bl_download_series grabs a whole cycle in one call
• Format conversion — convert between epub, fb2, txt, pdf with full preservation of chapters, bold/italic, inline images, cover and metadata; zipped books (.fb2.zip) are handled automatically; bl_convert_batch converts an entire directory at once
• Reading stats — yearly progress, top authors/genres, average rating
• Reading challenge — set yearly goals and track progress
• Import/Export — bulk import from text, export to JSON/CSV/Markdown

Installation

No installation needed if you use uvx — it fetches the package automatically.

Or install manually (requires Python 3.11+):

python3.11 -m venv .venv
source .venv/bin/activate
pip install mcp-book-lover

Connecting to MCP clients

Claude Code

claude mcp add book-lover uvx mcp-book-lover

To make it available in all projects, add to ~/.claude/mcp.json:

{
  "mcpServers": {
    "book-lover": {
      "command": "uvx",
      "args": ["mcp-book-lover"]
    }
  }
}

Claude Desktop / Kiro / Cursor

Add to the MCP config file of your client:

{
  "mcpServers": {
    "book-lover": {
      "command": "uvx",
      "args": ["mcp-book-lover"]
    }
  }
}

Development

git clone https://github.com/MIt9/mcp-book-lover
cd mcp-book-lover
uv sync

Test interactively with the MCP inspector:

mcp dev src/mcp_book_lover/server.py

Tools

Tool	Description
bl_add_book	Add a book to your library (with series/order support)
bl_list_books	List books with filters (status, author, genre)
bl_find_in_library	Search YOUR library by title, author, series, description
bl_update_book_status	Change reading status
bl_delete_book	Delete a book with all its reviews and quotes
bl_list_series	List all series with progress
bl_series_books	Show books in a series with order and status
bl_review_book	Write a review + rating (1–5)
bl_get_reviews	View reviews for a book
bl_review_series	Write a review for a book series
bl_get_series_reviews	View series reviews
bl_save_recommendation	Save a recommendation for future reading
bl_list_recommendations	List saved recommendations
bl_delete_recommendation	Remove a recommendation from the list
bl_add_quote	Save a quote from a book
bl_list_quotes	View saved quotes
bl_get_recommendations	Get recommendations by author/genre via Open Library
bl_search_books	Search online across 8 sources
bl_search_series	Find ALL books of a cycle/series — groups numbered parts (e.g. "Брутфорс 1..6") into one result
bl_suggest_next	What to read next (scored by ratings, series reviews, fav authors/genres)
bl_find_download	Find download links (LibGen, Flibusta)
bl_download_book	Download a book from searchfloor.org to ~/Books
bl_download_series	Download ALL books of a cycle/series in one call
bl_convert_book	Convert book file — preserves chapters, formatting, images; auto-extracts .fb2.zip/.epub.zip
bl_convert_batch	Convert every book file in a directory to one format (great for a downloaded cycle)
bl_reading_stats	Reading statistics
bl_set_goal	Set a yearly reading challenge
bl_goal_progress	Check reading challenge progress
bl_import_list	Import books from text list
bl_export_library	Export library (json/csv/markdown)

Book Statuses

Status	Meaning
want_to_read	On the wishlist, don't have the file yet
downloaded	File ready, queued for reading
reading	Currently reading/listening
finished	Done

Format Conversion

All conversions preserve chapter structure, metadata (title, author, language), cover image, bold/italic text, and inline images.

From \ To	epub	fb2	pdf	txt
fb2	✅	—	✅	✅
epub	—	✅	✅	✅
txt	✅	✅	✅	—
pdf	✅	✅	—	✅

PDF output uses a system Unicode font (Arial on macOS, DejaVu on Linux) for Cyrillic support.

Zipped book files (.fb2.zip, .epub.zip) are extracted automatically before parsing, so you can convert them directly.

Typical cycle workflow

# 1. Find every book of the cycle at once
bl_search_series("Брутфорс")

# 2. Download all parts in one call
bl_download_series("Брутфорс", author="Иван Катиш")

# 3. Convert the whole directory (auto-extracts .fb2.zip)
bl_convert_batch("~/Books", "epub")

Search Sources

Sources are auto-selected based on query language (Cyrillic → uk/ru sources, Latin → en sources):

Source	Languages	Type
Google Books	en, uk, ru	API
Open Library	en, uk, ru	API
Author.Today	uk, ru	Scraping
Knigogo	uk, ru	Scraping
Searchfloor	uk, ru	Scraping + download
LibGen	en, uk, ru	Scraping
Flibusta	ru	Scraping
Project Gutenberg	en	Scraping

Storage

SQLite database at ~/.mcp-book-lover/books.db. Downloaded books go to ~/Books by default.

Requirements

• Python ≥ 3.10
• No external services or API keys required
• System Unicode font (Arial/DejaVu) for PDF output with Cyrillic

License

MIT