How karwap22/ragi works
Project type and purpose
• This is a full-stack browser extension (“AI Buddy”) that combines a frontend (the extension UI + background script + content scripts) with a Python/Flask backend.
• Purpose: extract page text → send it to a retrieval-augmented generation (RAG) pipeline → generate streaming summaries and Q&A responses → display and persist results per URL.
File structure
├─ backend/
│ ├─ main.py # entry point – starts Flask server & SSE endpoints
│ ├─ server.py # defines REST/SSE routes for summarization & Q&A
│ ├─ utils.py # text splitting, embedding generation, FAISS setup
│ └─ requirements.txt
├─ frontend/
│ ├─ manifest.json & manifest_sidebar.json # extension metadata, permissions, scripts
│ ├─ background.js # service worker handling network calls & message routing
│ ├─ popup.html/css/js # UI for “Extract Text”, display summary & Q&A form
│ ├─ question.js # logic for sending questions & rendering streaming answers
│ └─ summary.js # logic for initiating summarization and handling SSE stream
└─ README.md, Demo_compressed.mp4, .gitignore
Key README insights
• Tech stack: JavaScript/HTML/CSS (extension) + Python/Flask + Flask-CORS + LangChain + Ollama LLM + FAISS.
• Workflow summary: frontend → backend RAG (embeddings + vector search + LLM) → SSE streaming back.
• Persistent storage: local extension storage for summaries/Q&A per URL.