A Model Context Protocol (MCP) server for searching and reading books from your Calibre ebook library.
- Full-text search with automatic phrase matching and fuzzy fallback
- Search by author or title with partial matching
- Extract text excerpts from books with keyword highlighting and pagination
- Pure bash implementation using
calibredbCLI - Returns Calibre deep links (
calibre://) and file URLs (file://) - Robust timeout handling for macOS compatibility
- Calibre installed at
/Applications/calibre.app - A Calibre library at
~/Calibre Library - Text (.txt) exports of your books for excerpt functionality
- Bash 4.0 or later
- Clone this repository:
git clone https://github.com/trieloff/calibre-mcp.git
cd calibre-mcp- Make the script executable:
chmod +x calibre-mcp.shTest the server using the MCP Inspector:
npx @modelcontextprotocol/inspector /path/to/calibre-mcp.shAdd to your Claude Desktop configuration:
{
"mcpServers": {
"calibre": {
"command": "/path/to/calibre-mcp.sh"
}
}
}Search books using Calibre's full-text search engine.
Parameters:
query(required): Search query. Multi-word queries are automatically phrase-searchedlimit(optional): Maximum results (default: 50)fuzzy_fallback(optional): Fallback query with related terms if exact search fails
Example:
{
"query": "machine learning",
"fuzzy_fallback": "AI artificial intelligence ML neural networks deep learning"
}Search books by author name with partial matching.
Parameters:
author(required): Author name (partial match supported)limit(optional): Maximum results (default: 50)
Search books by title with partial matching.
Parameters:
title(required): Book title (partial match supported)limit(optional): Maximum results (default: 50)
Extract text excerpts from books with keyword context.
Parameters:
book_id(required): Calibre book IDkeyword(optional): Search term to find in the bookcontext_lines(optional): Lines of context around matches (default: 10)max_results(optional): Results per page (default: 10)page(optional): Page number for pagination (default: 1)
Search results include:
- Book metadata (title, authors, series, tags, publisher)
- Calibre deep link for opening in Calibre
- File URL for direct EPUB access
- Available formats
- Whether text format is available for excerpts
The server logs all requests and responses to requests.log for debugging.
- Requires Calibre to be closed or may experience timeouts
- Full-text search requires indexed books
- Excerpt extraction only works with .txt exports
Apache License 2.0 - See LICENSE file for details.