# Shopping‑List AI Assistant

This repository contains a minimal example of a hierarchical LangChain agent that can estimate prices for products in a city.  The top‑level agent uses the ``get_price`` tool, which internally creates a short‑lived sub‑agent to generate a markdown table row with the price.

## Project structure

- **main.py** – entry point; demonstrates three example calls and prints the full message chain.
- **agent.py** – defines the ``get_price`` tool used by the main agent.
- **requirements.txt** – Python dependencies (no pinned versions, only ``>=``).
- **README.md** – this documentation.

## Installation

```bash
pip install -r requirements.txt
# Ensure LM Studio is running at http://localhost:1234/v1 or set the env vars below:
export LOCAL_LLM_MODEL=gpt-3.5-turbo
export LOCAL_LLM_BASE_URL=http://localhost:1234/v1
export LOCAL_LLM_API_KEY=fake
```

## Running the examples

```bash
python main.py
```

You should see three blocks of output, each showing the assistant’s messages and the tool calls that were made.

## How it works

1. **Sub‑agent** – The ``get_price`` tool creates a lightweight sub‑agent with a focused system prompt to generate a single markdown table row containing product name, price (rub.) and store.
2. **Main agent** – Uses the ``get_price`` tool to estimate prices for each product in the user’s list.
3. **Output** – The assistant prints all intermediate messages (tool calls) followed by the final answer summarizing the shopping list and total cost.

## Dependencies

- `langchain-openai>=0.3.0`
- `langgraph>=0.2.0`
- `langchain-core>=0.3.0`
- `python-dotenv>=1.0.0`

All versions are specified with ``>=`` to avoid pinning to potentially non‑existent releases.