MCP Server Ragdocs

🚀 MCP-Server-RAG Document Management Guide

The MCP-Server is a document management system based on Retrieval Augmented Generation (RAG) technology. It offers various tools to help you process, search, and manage documents. Here are the main features of this server:

✨ Features

• Document Storage: Supports uploading and storing documents in multiple formats (e.g., PDF, Word, PPT, etc.).
• Full-Text Search: Quickly find relevant document content through natural language queries.
• URL Crawling: Extract links from specified web pages and analyze their content.
• Queue Management: Process documents to be indexed and monitor the processing progress.

📦 Installation

Install with Docker

docker pull sanderkooger/mcp-server-ragdocs:latest

Local Installation

1. Clone the repository:

   git clone https://github.com/sanderkooger/mcp-server-ragdocs.git
   cd mcp-server-ragdocs
   

2. Install dependencies:

   npm install
   

📚 Documentation

Environment Variable Configuration

Before running the server, please set the following environment variables:

Log Configuration

Logs are output to the console by default. You can also specify other storage locations by modifying the configuration file. The log format follows the standard structured log specification.

🔧 Technical Details

The server is designed with a modular architecture, and each component has a clear separation of responsibilities. For a detailed architecture diagram, please refer to ARCHITECTURE.md.

💻 Usage Examples

Search Documents

Use natural language queries to search for stored documents and return relevant paragraphs and context.

Input Parameters:

• query (string): The text to search for, which can be natural language, specific terms, or code snippets.
• limit (number, optional): The maximum number of results to return (1 - 20, default value is 5).

List Sources

List information about all currently stored document sources, including URLs, titles, and update times.

Extract Links

Extract and analyze all URLs from a specified web page. Supports automatically adding the extracted URLs to the processing queue.

Input Parameters:

• url (string): The full web page address to crawl.
• add_to_queue (boolean, optional): Whether to add the extracted URLs to the queue, default is false.

Delete Documents

Permanently delete specified documents based on URLs. Each URL must be exactly the same as when it was stored.

Input Parameters:

• urls (array of strings): The list of URLs to delete.

List Queue

Display information about all currently pending document URLs, including status and processing times.

Run Queue

Process documents in the queue one by one and index their content into the database. Provides detailed processing progress feedback.

Empty Queue

Remove all pending URLs from the queue. This operation is irreversible, so use it with caution.

🔧 Technical Details

This package follows the principle of modular design, and each part has a clear function. For the specific file and directory structure, please refer to PROJECT_STRUCTURE.md.

📄 License

This project is open-sourced under the MIT License. Please view LICENSE.md for the specific license content.

With the above guide, you can quickly start using the MCP-Server to manage your document resources. If you need further information, please feel free to consult the detailed documentation or contact the development team.