Entity - A Cross-Platform LLM Agent

Entity is a Python-based AI agent designed to interact with your operating system, execute commands, and control programs across Windows, macOS, and Linux.

Architecture

This agent uses a local Large Language Model (LLM) powered by Ollama. By default, it connects to a locally running Ollama server and uses the default model (e.g., llama3). All data and interactions remain private and on your local machine.

The core components are:

• Agent Logic: The main Python application that orchestrates tasks.
• LLM Service: An Ollama server running a local model (e.g., Llama 3).
• Platform Interaction: Modules for interacting with the specific operating system's terminal and applications.

Getting Started

Prerequisites

• Python 3.8+
• Ollama installed and running (default: ollama serve)
• A downloaded Ollama model (default: llama3, run ollama run llama3 to download)

Installation

From PyPI (recommended)

pip install entityagent

From source

git clone https://github.com/prakashsellathurai/entity
cd entity
pip install .

Running the Agent (Default Ollama)

You can run the agent from the command line (it will use the default Ollama model):

entity-agent

or

python -m entityAgent.runtime

TODO / Roadmap

• Allow configuration of LLM model (e.g., choose llama3, mistral, etc.)
• Add support for configuring LLM server URL/port
• Add UI options (CLI, Web, GUI)
• Add configuration file for user preferences
• Extend platform interaction capabilities

Usage

You can interact with the agent in two ways:

1. Natural Language:

   > Tell me a joke.
   > What is the capital of France?
   

2. Commands:

   • Execute a shell command:

     > run: ls -l
     

   • List running processes:

     > run: list_processes
     

Testing

To run the test suite, execute the following command:

pytest