# RAG Agent

## Overview

This project implements a Retrieval‑Augmented Generation (RAG) agent that can search a local knowledge base stored in ChromaDB and the web via Tavily. The agent automatically chooses the appropriate source and indicates it in the response.

## Features

- Local vector store with Ollama embeddings (`nomic-embed-text`)
- Web search powered by Tavily
- Two tools: **Local KB Search** and **Web Search**
- Automatic source selection
- Persistent vector store between runs
- CLI chat loop with exit command

## Installation

```bash
# Pull required models
ollama pull llama3
ollama pull nomic-embed-text

# Install Python dependencies
pip install langchain langchain-chroma langchain-tavily langchain-ollama tavily-python chromadb python-dotenv
```

## Setup

Create a `.env` file in the project root with your Tavily API key:

```
TAVILY_API_KEY=your_api_key_here
CHAT_BASE_URL=http://localhost:11434/v1
CHAT_API_KEY=ollama
CHAT_MODEL=llama3
```

## Usage

```bash
python main.py --docs_dir path/to/documents
```

- `--docs_dir` (optional) – Directory containing `.txt` or `.md` files to index into the vector store. If omitted, the agent will use the existing persisted store.

### Example

```
You: What are the latest news about AI agents?
Agent: 1. AI Agents in 2024 (https://example.com) ...
Source: tavily

You: Tell me about LangGraph in my notes.
Agent: LangGraph is a framework for building ...
Source: chromadb
```

## Exiting

Type `exit` or `quit` to exit the chat loop.

## License

MIT