home / mcp / calibre rag mcp server

Calibre RAG MCP Server

Calibre RAG MCP Server - Enhanced MCP server for Calibre library with RAG capabilities

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "ispyridis-calibre-rag-mcp-nodejs": {
      "command": "node",
      "args": [
        "path/to/calibre-rag-mcp-nodejs/server.js"
      ]
    }
  }
}

You run an MCP server that adds retrieval-augmented search capabilities to your Calibre library. It vectorizes books, supports multiple formats, and enables project-based contexts for fast, accurate answers from your local library data.

How to use

Launch and interact with the Calibre RAG MCP Server through your MCP client to perform semantic searches across your ebook collection, fetch specific passages, and manage RAG projects. Create projects for different contexts, add books to those projects for vectorization, and run searches within a project to retrieve contextually relevant passages.

How to install

Prerequisites you need before installation include Node.js and Calibre on Windows, along with optional image and OCR tools for enhanced processing.

# Prerequisites
node -v  # should be >= 16.0.0
calibre --version

# Optional (enhanced features)
magick -version
`tesseract --version`

# Install and set up
git clone https://github.com/yourusername/calibre-rag-mcp-nodejs.git
cd calibre-rag-mcp-nodejs
npm install

Additional configuration notes

The server detects your Calibre library location automatically. If you need custom settings, update the configuration in the main server script to tailor paths, project defaults, and processing options.

Project structure and starting the server

The project stores RAG projects under the projects directory. To start the local MCP server, run the start command from your terminal.

npm start

MCP configuration for a local stdio server

If you want to expose this server as an MCP endpoint via a local runtime, include the following configuration in your MCP client. It runs the server locally using Node and points to the server script.

{
  "mcpServers": {
    "calibre_rag": {
      "command": "node",
      "args": ["path/to/calibre-rag-mcp-nodejs/server.js"]
    }
  }
}

Available tools

search

Semantic search across your ebook library using vector representations for context-aware results.

fetch

Retrieve specific content from books based on context or queries.

list_projects

List all existing RAG projects to switch contexts quickly.

create_project

Create a new RAG project to isolate a set of books or topics.

add_books_to_project

Add books to a chosen project so they can be vectorized and indexed.

search_project_context

Search within the contents and context of a specific project for targeted results.