nturumel/flask

TWK-Backend

This is the backend for the TWK project, a Flask application that uses a cache of chat agents. Each chat agent is associated with a session ID.

Setup

The project uses Poetry for dependency management. You will need to have Poetry installed to set up and run the project.

To install the project dependencies, navigate to the project directory and run:

poetry install

This will install the following dependencies:

• Flask: ^2.3.2
• cachetools: ^5.3.1
• python-env-loader: ^0.0.2
• python-dotenv: ^1.0.0
• openai: ^0.27.8
• tiktoken: ^0.4.0

And the following dev dependencies:

• pytest: ^7.3.2
• flake8: ^6.0.0
• black: ^23.3.0
• mypy: ^1.3.0
• sphinx: ^7.0.1
• bandit: ^1.7.5

Running the application

Once the dependencies are installed, you can start the application with the following command:

FLASK_APP=main

flask run

Windows

set FLASK_APP=main
flask run

This will start the Flask server and the application will be accessible at http://localhost:5000.

API

The application exposes two endpoints:

• /initialiseAgent: This endpoint accepts a POST request with a JSON body. It initializes a chat agent and associates it with the provided session ID.
• /chatWithAgent: This endpoint also accepts a POST request with a JSON body. It uses the provided session ID to retrieve a chat agent from the cache and returns a response from the agent to the provided query.