Keyword RAG FAQs

Question 1

Can I customize the search parameters for Keyword RAG?

Accepted Answer

Yes, Keyword RAG provides configurable parameters for the BM25 algorithm (k1, b) and document chunking (minWords, contextWindowSize). This allows you to fine-tune search relevance and performance based on your specific document characteristics and needs.

Question 2

What is Keyword RAG?

Accepted Answer

Keyword RAG is an MCP (Model Context Protocol) server designed for efficient knowledge retrieval from Markdown (.md) and MDX (.mdx) documents. It utilizes the BM25 algorithm to provide relevant search results from your structured documentation.

Question 3

How does Keyword RAG integrate with Claude Desktop?

Accepted Answer

It integrates seamlessly as an MCP server, allowing Claude Desktop users to leverage Keyword RAG's search, retrieval, and domain listing API tools. This enhances Claude's ability to access and utilize your internal knowledge base for richer interactions.

Question 4

Is Keyword RAG difficult to set up?

Accepted Answer

Not at all. Keyword RAG offers simple one-command installation scripts for macOS/Linux and Windows, which automate setup, dependency installation, project building, and even create example files and configuration for immediate use.

Question 5

What types of documents can Keyword RAG process?

Accepted Answer

Keyword RAG supports Markdown (.md), MDX (.mdx), and .markdown file formats. You can organize your documents into configurable domains (e.g., company, product, technical) for targeted retrieval.

Keyword RAG

概要

主な機能

ユースケース