RAG Chatbot Template

Local RAG chatbot template for documents, source-aware retrieval, and optional FastAPI endpoints.

The system retrieves relevant document chunks and answers with source-aware context. It does not claim perfect answers; it makes the retrieved evidence visible.

Features

• ingest .txt and .pdf files
• chunk documents into overlapping text windows
• build a local TF-IDF vector store
• retrieve relevant chunks for a question
• generate a source-aware answer context
• return retrieved chunk ids, source paths, scores, and previews
• store chunking metadata with the local vector store
• expose /ask and /health with FastAPI
• tested without external API calls

Demo

pip install -e ".[dev]"
rag-ingest documents --store vector_store/store.json
rag-chat "What does the refund policy say?" --store vector_store/store.json
uvicorn rag_chatbot_template.app:app --reload

API

curl -X POST http://127.0.0.1:8000/ask \
  -H "Content-Type: application/json" \
  -d "{\"question\":\"What does the support policy say?\"}"

Sample Delivery Report

Run Tests

ruff check .
pytest

Scope

This template works with user-provided documents. For production, add authentication, monitoring, document refresh jobs, and a hosted vector database if the corpus grows.