MCP Book Lover

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.

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

Book Statuses

Format Conversion

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

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):

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