LLM AGENT

LLM AGENT

A CLI-based AI agent powered by Google's Gemini.


Report Bug · Request Feature

Table of Contents

1. About The Project
   • Built With
2. Getting Started
   • Prerequisites
   • Installation
3. Usage
4. Customization Guide
5. Acknowledgments

About

This was my first working version—it's messy and buggy, but it helped me understand the fundamentals. Feel free to check it out, but version 2 will be a much cleaner, improved version.

Built With

Getting Started

To get started with the app, follow these simple steps:

Prerequisites

List things needed to use the software and how to install them:

• Node.js
• Google Gemini API key

Installation

1. Clone the repository:

git clone https://github.com/aditsuru-git/llm-agent-v1.git

2. Install dependencies:

npm install
npx playwright install

3. Edit .env and add your Google Gemini API key:

API_KEY=your_gemini_api_key

4. Delete all the .gitkeep files from playground

Usage

Start the chat application:

npm run start

Features

• View images, videos, and listen to audio files
• Take screenshots of any webpage
• Perform CRUD operations on files and folders

AI can only access files & folders inside a root directory specified in src/config/configPath.js

Chat Commands

• sayonara() - Exit the chat
• reset() - Clear chat history and start fresh

Customization Guide

Overview

This chatbot can be customized to fit your needs, from appearance to personality and file access. Below are the key customization options.

UI Customization

Modify src/config/styles.json to change the chat interface colors:

{
	"welcome_color": "#ff8400",
	"instruction_color": "#828181",
	"user_color": "#00fff2",
	"chat_color": "#F0F8FF",
	"ai_color": "#ff0000"
}

Update these values to your preferred hex colors.

Personality Customization

Edit src/config/persona.js to adjust the AI's persona, speech style, and behavior.

For example, to modify how the AI interacts, tweak the persona constant:

export const persona = `
1: Identity:
You are [Your AI Name]...
`;

• Make changes to the character's traits, tone, or role as needed.
• Note: Be sure to update the chat history as well to ensure the persona changes take full effect.

Playground Path Customization

Change the AI's accessible root directory in src/config/configPath.js by modifying:

export const playGroundPath = path.join(__dirname, "..", "..", "playground");

Update this path to point to a different directory.

User Information Customization

Update src/config/userInfo.js to personalize the chatbot's knowledge of the user:

export const userInfo = `I'm [Your Name], a [Your Role]...`;

This allows the AI to tailor responses based on user details.

Chat History Management

Modify src/config/chatHistory.js to control past conversations. This enables the chatbot to adhere more accurately to the persona.

Example format:

export const chatHistory = [
	{ role: "user", parts: [{ text: "Hello!" }] },
	{ role: "model", parts: [{ text: "Hey there!" }] },
];

Clear or edit history as needed.

Additional Configuration

• Utility functions: Located in src/utils/ (e.g., fileOperations.js, validate.js).
• AI behavior rules: Managed in src/config/systemInstructions.js.

Customize these as needed to refine the chatbot's functionality.

License

Distributed under the MIT License. See LICENSE for more information.

Acknowledgments

• Gemini SDK Documentation

(back to top)