RAG Gateway

About

This tool acts as an MCP server, serving as a gateway between AI clients like Claude or ChatGPT and a RAG Knowledge Base Platform. It facilitates querying knowledge bases, retrieving chunks, listing knowledge bases and documents, and managing retrieval settings. The gateway enforces OAuth access tokens (JWT) on all MCP calls, ensuring secure communication, and handles proxying requests to the RAG backend. It also publishes OAuth metadata for client discovery and supports advanced retrieval features like Windowed Retrieval.

Key Features

• Automates RAG backend authentication with API keys or bearer tokens, including auto-refresh.
• Exposes MCP tools for querying knowledge bases, retrieving context, and managing KB settings.
• Supports advanced retrieval options including Windowed Retrieval (Context Expansion).
• Enforces OAuth access tokens (JWT) on all MCP calls for secure communication.
• 0 GitHub stars
• Publishes OAuth protected resource and authorization server metadata for client discovery.

Use Cases

• Integrating RAG Knowledge Base Platforms with AI clients like Claude or ChatGPT for enhanced conversational AI.
• Securing access to RAG knowledge bases through OAuth/OIDC for client applications.
• Providing a standardized MCP interface for diverse RAG operations, from querying to configuration management.