Summary

Agency is a python library that provides an Actor model framework for creating agent-integrated systems.

The library provides an easy to use API that enables you to connect agents with traditional software systems in a flexible and scalable way, allowing you to develop any architecture you need.

Agency's goal is to enable developers to create custom agent-based applications by providing a minimal foundation to both experiment and build upon. So if you're looking to build a custom agent system of your own, Agency might be for you.

Features

Easy to use API

• Straightforward class/method based agent and action definition
• Up to date documentation and examples for reference

Performance and Scalability

• Supports multiprocessing and multithreading for concurrency
• AMQP support for networked agent systems

Observability and Control

• Action and lifecycle callbacks
• Access policies and permission callbacks
• Detailed logging

Demo application available at examples/demo

• Multiple agent examples for experimentation
  • Two OpenAI agent examples
  • HuggingFace transformers agent example
  • Operating system access
• Includes Gradio UI
• Docker configuration for reference and development

API Overview

In Agency, all entities are represented as instances of the Agent class. This includes all AI-driven agents, software interfaces, or human users that may communicate as part of your application.

All agents may expose "actions" that other agents can discover and invoke at run time. An example of a simple agent could be:

class CalculatorAgent(Agent):
    @action
    def add(a, b):
        return a + b

This defines an agent with a single action: add. Other agents will be able to call this method by sending a message to an instance of CalculatorAgent and specifying the add action. For example:

other_agent.send({
    'to': 'CalcAgent',
    'action': {
        'name': 'add',
        'args': {
            'a': 1,
            'b': 2,
        }
    },
})

Actions may specify an access policy, allowing you to control access for safety.

@action(access_policy=ACCESS_PERMITTED) # This allows the action at any time
def add(a, b):
    ...

@action(access_policy=ACCESS_REQUESTED) # This requires review before the action
def add(a, b):
    ...

Agents may also define callbacks for various purposes:

class CalculatorAgent(Agent):
    ...
    def before_action(self, message: dict):
        """Called before an action is attempted"""

    def after_action(self, message: dict, return_value: str, error: str):
        """Called after an action is attempted"""

    def after_add(self):
        """Called after the agent is added to a space and may begin communicating"""

    def before_remove(self):
        """Called before the agent is removed from the space"""

A Space is how you connect your agents together. An agent cannot communicate with others until it is added to a common Space.

There are two included Space implementations to choose from:

• LocalSpace - which connects agents within the same application.
• AMQPSpace - which connects agents across a network using an AMQP server like RabbitMQ.

Finally, here is a simple example of creating a LocalSpace and adding two agents to it.

space = LocalSpace()
space.add(CalculatorAgent, "CalcAgent")
space.add(MyAgent, "MyAgent")
# The agents above can now communicate

These are just the basic features that Agency provides. For more information please see the help site.

Install

pip install agency

or

poetry add agency

The Demo Application

The demo application is maintained as an experimental development environment and a showcase for library features. It includes multiple agent examples which may communicate w