# RAG Agent with ChromaDB and Web Search

This repository implements a simple RAG (Retrieval‑Augmented Generation) agent that can answer questions using a local knowledge base stored in **ChromaDB** and also perform real‑time web search via **Tavily**. The agent automatically decides which source to use and reports the chosen source in the answer.

## Features

- **Local knowledge base** – Text files (.txt, .md) are loaded, chunked, and stored in a persistent ChromaDB collection.
- **Semantic search** – Uses Ollama embeddings (`nomic-embed-text`).
- **Web search** – Powered by Tavily.
- **Automatic source selection** – The agent chooses between local and web search based on the query.
- **CLI** – Simple chat loop with `exit` to quit.

## Setup

```bash
# 1. Create a virtual environment (optional but recommended)
python -m venv venv
source venv/bin/activate   # Windows: venv\Scripts\activate

# 2. Install dependencies
pip install -r requirements.txt

# 3. Pull required Ollama models
ollama pull llama3
ollama pull nomic-embed-text

# 4. Set your Tavily API key
export TAVILY_API_KEY=YOUR_KEY   # Windows: set TAVILY_API_KEY=YOUR_KEY
```

## Usage

1. **Load documents** – Place your `.txt` or `.md` files in the `documents/` folder.
2. **Run the agent**
   ```bash
   python agent.py
   ```
3. **Chat** – Type your question. Type `exit` to quit.

## Example

```
Query: Какие последние новости про AI-агентов?
[Web Search] ...
Источник: tavily

Query: Что в наших конспектах про LangGraph?
[Local KB] ...
Источник: chromadb
```

## Project Structure

- `vectorstore.py` – Functions to create and load the ChromaDB vector store.
- `rag_tools.py` – Two LangChain tools: `search_local_kb` and `web_search`.
- `agent.py` – Main script that sets up the agent and runs the chat loop.
- `requirements.txt` – Python dependencies.
- `README.md` – This file.

## License

MIT License.