This project is a microservices-based web application that allows you to:

• 🕸️ Scrape and parse website content

• 📝 Convert it into Markdown

• 🧬 Generate vector embeddings using OpenAI

• 🧠 Store them in ChromaDB

• 💬 Query them using an LLM (OpenAI or Gemini)

• 🖥️ Interact via a frontend

• 🔄 Dynamically add more content: While you're chatting, you can initiate additional scrapes to download and embed more information into the database without restarting the service.

git clone https://github.com/your-user/llm-chat-app.git
cd llm-chat-app

Create a .env file at the root of the project with the following content:

OPENAI_API_KEY=your_openai_key
GEMINI_API_KEY=your_gemini_key
MAX_FILES=number_of_files_to_download

⚠️ Make sure these variables are accessible by Docker Compose when running the containers.

Alternatively, you can export them manually:

export OPENAI_API_KEY=your_openai_key
export GEMINI_API_KEY=your_gemini_key

Make sure Docker is running, then launch everything:

docker compose up --build

This will spin up:

• ChromaDB on http://localhost:8000
• Scraper/Embedding API on http://localhost:8080
• LLM API on http://localhost:5050
• Frontend on http://localhost:3000

You can enrich LLM responses with additional structured context using a dictionary file:

• Place your custom dictionary.json inside llm/data/

• This file should contain extra definitions, acronyms, or reference data

• The LLM backend will use this during the query process to improve accuracy and context

.
├── LICENSE
├── README.md
├── assets/
│   └── llm-chat.png
├── compose.yml
├── embeddings/
│   ├── Dockerfile
│   ├── app.py
│   ├── requirements.txt
│   └── utils/
│       ├── createEmbeddingsDB.py
│       ├── extractWebInfo.py
│       └── scraper.py
├── frontend/
│   ├── Dockerfile
│   ├── app.js
│   ├── index.html
│   └── style.css
└── llm/
    ├── Dockerfile
    ├── app.py
    ├── data/
    │   ├── dictionary.json
    │   └── config.json
    ├── requirements.txt
    └── utils/
        └── llm.py

• ✅ Make sure .env is not accidentally ignored or overwritten

• 🔄 Restart containers if you change .env or environment variables

• 📁 Embeddings are saved inside the ChromaDB persistent volume under ./chromadb/

GNU GENERAL PUBLIC License — see LICENSE for details.