Komga Ebook Library

Search and read your Komga ebook collection through AI using Model Context Protocol

Connect your Komga digital library to AI assistants like Claude, enabling intelligent book discovery, content reading, and full-text search across your ebook collection.

🎯 Purpose

Bridge AI assistants with digital library infrastructure to provide authoritative source access for research, fact-checking, and academic citation. Core philosophy: Komga finds books, we search IN books.

✨ Key Features

• Smart Format Detection: Automatic EPUB vs PDF handling
• Full-Text Search: FlexSearch integration for instant content discovery
• Claude Vision OCR: Intelligent PDF processing for non-text content
• Session Caching: Optimized performance for repeated access
• Conservative Fallbacks: Graceful degradation prevents poor user experience

📚 Format Support Matrix

🚀 Quick Start

Installation

# Via NPM (recommended)
npx @sylweriusz/komga-book-search

# Via Smithery (for Claude Desktop)
npx @smithery/cli install @sylweriusz/komga-book-search --client claude

Configuration

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "komga-ebook-search": {
      "command": "npx",
      "args": ["@sylweriusz/komga-book-search"],
      "env": {
        "KOMGA_URL": "http://your-komga-server:25600",
        "KOMGA_USERNAME": "your-username",
        "KOMGA_PASSWORD": "your-password",
        "KOMGA_LIBRARY": "your-library-filter"
      }
    }
  }
}

🛠️ MCP Tools

search_library_books(query, library_filter?)

Discover books across library collections with format detection.

// Find academic books on specific topics
{
  "query": "artificial intelligence",
  "library_filter": "academic"  // optional
}

get_book_overview(book_id)

Retrieve complete book metadata including page count and format information.

{
  "book_id": "0K79C4PQAEH7J"
}

read_book_pages(book_id, start_page, end_page)

Access book content with intelligent format routing (max 15 pages).

• EPUBs: Returns clean text with chapter delimiters
• PDFs: Returns compressed images optimized for Claude Vision OCR

search_within_book(book_id, search_term)

Full-text search within individual books.

• EPUBs & Text PDFs: FlexSearch with context and highlights
• Image PDFs: Guidance to use read_book_pages with Claude Vision

💡 Use Cases

• Academic Research: Search scholarly books for citations and references
• Fact Verification: Cross-reference claims against authoritative sources
• Language Learning: Find conversation examples and vocabulary in context
• Technical Documentation: Navigate complex technical manuals efficiently

🏗️ Architecture

Built with Lean Architectural Methodology (LAM):

• Vertical Feature Slicing: Each MCP tool as self-contained feature
• Dependency Injection: Constructor-based, no singletons
• Strict Type Safety: Zero any types, comprehensive TypeScript
• Single Responsibility: One file, one purpose
• No Circular Dependencies: Clear dependency direction

📋 Requirements

• Node.js: 18.0.0 or higher
• Komga Server: Running instance with API access
• TypeScript: Included in dependencies

🔧 Development

# Clone and setup
git clone https://github.com/sylweriusz/komga-ebook-search.git
cd komga-ebook-search
npm install

# Build and run
npm run build
npm start

# Development mode
npm run dev

📝 Configuration Variables

🐛 Troubleshooting

Connection Issues

• Verify Komga server is accessible
• Check authentication credentials
• Ensure user has PAGE_STREAMING role

Search Problems

• Confirm KOMGA_LIBRARY filter matches existing libraries
• Check book format support (EPUB preferred for text search)

Performance Issues

• EPUBs index on first access (2-4 seconds)
• Subsequent searches use cached indexes (<1 second)
• Large PDFs may require patience for initial processing

📄 License

MIT © sylweriusz

🤝 Contributing

Contributions welcome! Please read our Contributing Guide and follow the LAM architectural principles.

🔗 Related Projects

• Komga - Comic/ebook server and web reader
• Model Context Protocol - Open standard for AI tool integration
• Smithery - MCP server registry and marketplace

Built for researchers, academics, and digital library enthusiasts who need AI-powered access to their book collections.