This full-stack project allows you to ask intelligent questions about the content of any webpage. It features:
- β A FastAPI backend that scrapes, embeds, stores, and retrieves contextual data from web URLs.
- 𧩠A Chrome Extension that connects to this backend and allows users to interact with it directly from their browser.
- π Scrape and extract text from any webpage
- π§ Use HuggingFace embeddings to vectorize text
- π Store and query vectors using Qdrant
- π€ AI chatbot generates context-aware answers
- ποΈ Logs queries and responses to MongoDB
- FastAPI
- Qdrant
- HuggingFace Transformers
- MongoDB (via
motor) - BeautifulSoup for web scraping
| Method | Endpoint | Description |
|---|---|---|
GET |
/api/v1/health |
Check if the API is up |
POST |
/api/v1/process_url |
Process a webpage and store embeddings |
POST |
/api/v1/get_answer |
Ask a question and get an AI-generated answer |
POST |
/api/v1/delete_collection |
Clear Qdrant vector collection |
pip install -r requirements.txt
uvicorn app:app --reload-
π₯ Loads content from the current tab URL
-
π§ Asks questions and gets AI answers
-
π¬ Display interface using a popup
-
Uses Chrome APIs (tabs, cookies, storage, etc.)
-
Open any webpage in your browser
-
Click the Chrome extension and hit "Load URL"
-
Ask any question about the webpage content
-
Get a contextual answer from the backend AI
-
Set up the following in your .env file or system environment:
-
MONGODB_URI -
QDRANT_URLor local config -
HuggingFace modela details (in embeddings.py)