Website | Getting Started | Live Demo Video

This is the main source code repository for doc0, an AI-powered documentation chat interface that allows users to interact with technical documentation giving accurate results.

Why doc0?

• Smart Documentation Search: Leverages ChromaDB for semantic search across multiple framework documentations
• Multi-Framework Support: Currently supports Kestra, React, Next.js, Astro, and Redux documentation
• Accurate results: Give accurate results with latest documentation including source links
• Authentication Google OAuth integration with request limiting for free users

Did not won wemakedevs Hackathon

• Took all the servers downs there are costing too much
• But i was a good experience

Youtube demo video

Quick Start

1. Clone the repository
2. Follow the Installation Guide
3. Start developing with npm run dev for frontend and uvicorn main:app --reload for backend

Prerequisites

• Node.js (v18+)
• Python (v3.8+)
• Google OAuth credentials
• Gemini API key
• ChromaDB server

Installation

Frontend Setup

1. Clone & Navigate

   git clone https://github.com/sreehariX/doc0
   cd frontend

2. Install Dependencies

   npm install

3. Configure Environment Variables Create a .env file in the frontend directory:

   VITE_GOOGLE_CLIENT_ID=your_google_client_id

4. Start Development Server

   npm run dev

Backend Setup

1. Navigate to Backend Directory

   cd fastapi

2. Create & Activate Virtual Environment

   python -m venv venv
   source venv/bin/activate  # On Windows: .\venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

4. Configure Environment Variables Create a .env file in the backend directory:

   GEMINI_API_KEY=your_gemini_api_key
   CHROMA_SERVER_HOST=your_chromadb_server_url

5. Start Development Server

   uvicorn main:app --reload

Features

Documentation Search

• Semantic search powered by ChromaDB
• Support for multiple framework documentations
• Gives most accurate summary response

Authentication & Rate Limiting

• Google OAuth integration
• 10 free requests per day for non-authenticated users
• Unlimited requests for authenticated users until we run out of google credits

Modern UI Features

• Responsive design for all devices
• Code syntax highlighting

Tech Stack

Frontend

• React with TypeScript
• Tailwind CSS for styling
• Framer Motion for animations
• Zustand for state management

Backend

• FastAPI
• ChromaDB for vector search
• Google Gemini AI for response generation
• Python scrapers for documentation collection

Hosting

• Kestra is hosted on azure vm 16gb 2vcpu using docker
• Chromadb is also hosted on azure vm 16gb 2vcpu using docker
• Fastapi is hosted on azure vm 8gb 2vcpu using apache server for configuration
• Frontend vite+react+typescript is deployed on vercel

Kestra

• By using kestra we are automating the vectore embedding proecess
• Whenever i upload a file in azure blob storage it will trigger kestra to run a flow
• The flow execution will connect to chromadb server and make the chromadb embedded collection of the csv file

Documentation Scraping

The project includes custom scrapers for:

• React documentation
• Next.js documentation
• Redux documentation
• Astro documentation
• Kestra documentation

Each scraper out is csv file using that you can embed in vector database.

Contributing

We welcome contributions! Please see commit your messages properly

License

MIT License