Ramesh143-code/Vectorless-RAG-Document-QA — reverse-engineered prompt

Build me a simple document question answering app in Python with a basic web frontend. I want to be able to add or upload documents, ask normal questions in a chat style box, and get answers that are based only on the document content.

Please keep it lightweight and make it work without needing a separate vector database. If the app can search through the document text directly, pull out the most relevant parts, and send that context to an AI answer step, that’s perfect. Show the user which document or text snippet the answer came from so it doesn’t feel like it’s guessing.

Use the existing Python server files and frontend folder as the starting point. Make the app easy to run locally, with clear commands in the README. Add basic error handling for empty uploads, unsupported files, and questions that the documents don’t answer. Look up current docs online if you need to.