🖤 Igris AI Agent

An AI-powered real-time assistant built with LiveKit Agents and Google Gemini Realtime API. Igris can process text & audio in real-time and respond with natural speech.

---

🚀 Features

• 🎤 Real-time audio & text interaction
• ⚡ Powered by Gemini 2.0 Flash Experimental model
• 🗣️ Voice synthesis (Puck voice by default)
• 🔑 Secure API key management
• 🛠️ Runs inside a virtual environment for clean setup

---

📦 Installation

1. Clone the repo

git clone https://github.com/your-username/Igris.git
cd Igris

2. Create a virtual environment

python -m venv venv

3. Activate it

Windows (PowerShell):

.\venv\Scripts\Activate.ps1

Linux / macOS:

source venv/bin/activate

4. Install dependencies

pip install -r requirements.txt

---

🔑 API Key Setup

This agent requires a Google Gemini API key.

Option 1: Temporary (PowerShell)

$env:GOOGLE_API_KEY="YOUR_API_KEY_HERE"

Option 2: Permanent with .env (recommended)

1. Install dotenv:

   pip install python-dotenv

2. Create a .env file in the project root:

   GOOGLE_API_KEY=YOUR_API_KEY_HERE
   

3. The agent automatically loads .env.

---

▶️ Running the Agent

py agent.py console

Connect to playground

uv run agent.py dev

Console Mode

python agent.py console

Controls:

• [Ctrl+B] → Toggle between Text/Audio mode
• [Q] → Quit session

Development Mode

uv run agent.py dev

---

🛠️ Project Structure

Igris/
├── agent.py          # Main entrypoint
├── requirements.txt  # Python dependencies
├── .env              # API key (not committed to GitHub)
└── README.md         # This file

---

⚠️ Troubleshooting

• Error: "API key is required" → Ensure you set GOOGLE_API_KEY in your environment or .env

• Dependencies not found → Run pip install -r requirements.txt again

• PowerShell script execution blocked → Run:

  Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

---