AskWeb is a Chrome extension that lets users ask questions about the website they are currently visiting.
Instead of manually reading long articles, blogs, documentation pages, or wiki pages, users can simply ask questions in natural language and get answers based on the content of that webpage.
This project was built to understand how RAG (Retrieval-Augmented Generation) works in a real-world application.
AskWeb follows a simple flow:
- Extracts content from the current webpage
- Cleans the page content using Mozilla Readability
- Splits content into smaller chunks
- Converts text into vector embeddings
- Stores vectors in ChromaDB
- Retrieves the most relevant chunks based on the user's question
- Uses Gemini to generate an answer from the retrieved context
- React
- Tailwind CSS
- Mozilla Readability
- Chrome Extension APIs
- React Markdown
- FastAPI
- LangChain
- Google Gemini
- HuggingFace Embeddings (
all-MiniLM-L6-v2) - ChromaDB
- Ask questions about any webpage
- Reads and understands webpage content
- Filters unnecessary page noise
- Supports markdown-formatted answers
- Keeps chat history during current session
- Works on articles, blogs, wiki pages, and documentation
git clone https://github.com/Fluffy-debuger/AskWeb.git
cd AskWebGo to the backend folder:
cd WEBRAGInstall dependencies:
pip install -r requirements.txtCreate .env file:
GOOGLE_API_KEY= <Google_api_key>
HF_TOKEN = <HuggingFace_Token_key>Run backend server:
uvicorn main:app --reloadMove into frontend folder:
cd frontendInstall dependencies:
npm installBuild extension:
npm run build- Open Chrome
- Go to:
chrome://extensions/
- Enable Developer Mode
- Click Load unpacked
- Select your frontend build (ie.dist) folder (after runing npm run build)
Extension is now ready.
- Works best on text-heavy pages
- No persistent memory
- Vector DB resets on every request
- Persistent vector database
- Multi-page context memory
- Better content filtering
- PDF support
- Improved UI


