ace-twins-4-verts

This project is composed of two main components: an activity logger and a multi-tool agent. The ultimate goal is for the agent to use the data collected by the activity logger to answer questions about user activity.

Components

Activity Logger (activity_logging)

This is a Python script that runs on macOS to track the user's active application window and idle time.

• How it works: The script uses PyObjC to interface with the macOS window server and records the active application and window title at regular intervals.
• Output: The logger generates a new CSV file each day containing the activity data.
• Dependencies: pyobjc-core, pyobjc-framework-quartz

Multi-Tool Agent (agents/multi_tool_agent)

This is an agent built with the Google Agent Development Kit (ADK).

• Current Capabilities: The agent can currently answer questions about the time and weather.
• Future Goal: The agent will be extended to process the data from the activity logger to provide insights into user behavior and activity patterns.
• Dependencies: google-adk

Future Work

• Integrate the activity_logging output as a data source for the multi_tool_agent.
• Develop tools for the agent to parse, analyze, and answer questions about the activity log data.

Getting Started

Prerequisites

'''bash source venv/bin/activate pip install -r requirements.txt '''

Running Tests

To run the tests for the agent, use the following command:

PYTHONPATH=./agents adk eval agents/multi_tool_agent tests/evaluation.test.json

The python path makes the tools module visible.

Interacting with the Agent

To interact with the agent, you can use the adk web command:

PYTHONPATH=. adk web agents

This will start a web server with a user interface where you can chat with the agent.