Watsonx RAG Chatbot

Example Use Cases

• 1

  Answer user questions grounded in PDF manuals or domain documents by retrieving relevant text chunks and generating concise responses.

• 2

  A persistent, local vector store (ChromaDB) to speed up repeated document searches and avoid re-indexing on each run.

• 3

  Expose custom tooling to chat interfaces (e.g., Claude Desktop or other MCP-capable clients) through the Model Context Protocol.

Description

This repository provides a Python MCP server that extracts text from PDFs, builds a persistent ChromaDB vector index, and exposes a RAG tool which calls Watsonx.ai to generate answers grounded in document context. It includes a stdio-based MCP server, a sample client that launches the server and calls the tool, and an optional Flask frontend and Claude Desktop integration example. The design enables domain-specific question answering and easy integration into agent workflows via the Model Context Protocol.