LangGraph Query Router

This example demonstrates a multi-agent graph which utilizes an LLM within a query router to determine where to route a user query. Depending on the complexity of the user query, it will route to either a simple model or an advanced model.

The purpose of this repository is to demonstrate a simple graph with multiple agents & a conditional edge. It is by no means production ready, or capable of handling "real" requests, the prompts for the handle_advanced and handle_simple instruct the LLMs to pretend to answer the original query.

A visual representation of the graph

Requirements

1. Install Python (compatible version required, e.g., 3.x)
2. Install dependencies:

   pip install -r requirements.txt

3. Configure your environment:
   • Copy the example config file:

     cp ./config/config-example.yml ./config/config.yml

   • Add your API keys and update model configurations in config.yml.

Running the Example

To run the application:

1. Execute the main entry point:

   python app.py

2. The application uses two hardcoded queries:
   • One routed to the advanced model
   • Another routed to the simple model

Visualizing the Graph

Included in the src dir is a utility script generate_diagram.py which can be executed to generate a png visualizing the graph within this application. This script was used to generate the visualization above.

python generate_diagram.py

Notes

• This repository is intended for educational purposes and quick prototyping.
• See class level docstrings within ./core/language_models.py for limitations with the current language model manager.