Skip to content

Easy to use openAI framework for interacting with AI assistants

Notifications You must be signed in to change notification settings

AirspaceTechnologies/hatch

Repository files navigation

hatch : a framework for ai agents

Description

The purpose of this project is to facilitate interaction with different AI agents. These agents can perform various tasks, automate workflows, and aid in the development process.

Full control

There are zero built in system prompts.

Installation

Step-by-step instructions on how to install and set up the project.

  1. Clone the repository:
$ git clone https://github.com/michaelirey/hatch.git
  1. Navigate to the project directory:
$ cd hatch
  1. Install the required packages:
$ pip install -r requirements.txt
  1. Make sure openai api key is set
$ export OPENAI_API_KEY=...

Usage

Instructions and examples on how to use the project.

$ python app.py

AI Function Calling

The project allows you to interact with various AI agents. For example, in app.py, a HelloWorldAgent is used to look up domain information: See: https://platform.openai.com/docs/assistants/tools/function-calling

Thread Summary

The thread_summary.py script provides a summary of a specific thread. To use it, run the following command:

$ python thread_summary.py <thread_id>

This script uses the ThreadAnalyzer and ReportPrinter classes to analyze and print the thread summary.

Importing Assistants

You can import new assistants using the import_assistant.py script. This script retrieves the assistant configuration from OpenAI and generates the corresponding Python class.

To import an assistant, run:

$ python import_assistant.py <assistant_id>

This will save the assistant configuration and generate a new Python class in the assistants directory.

Tests

Instructions on how to run tests.

$ pytest

License

Information about the project's license. MIT License

About

Easy to use openAI framework for interacting with AI assistants

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published