LiveKit Agent with MCP Tools

A voice assistant application built using the LiveKit Agents framework, capable of using Multimodal Control Protocol (MCP) tools to interact with external services.

Features

• Voice-based interaction with a helpful AI assistant
• Integration with MCP tools from external servers
• Speech-to-text using Deepgram
• Natural language processing using OpenAI's GPT-4o
• Text-to-speech using OpenAI
• Voice activity detection using Silero

Prerequisites

• Python 3.9+
• API keys for OpenAI and Deepgram
• MCP server endpoint

Installation

1. Clone this repository:

   git clone https://github.com/livekit-examples/basic-mcp.git
   cd basic-mcp
   

2. Install the required packages:

   pip install -r requirements.txt
   

3. Create a .env file with your API keys and configuration:

   OPENAI_API_KEY=your_openai_api_key
   DEEPGRAM_API_KEY=your_deepgram_api_key
   ZAPIER_MCP_URL=your_mcp_server_url
   

Usage

Run the agent with the LiveKit CLI:

python agent.py console

The agent will connect to the specified LiveKit room and start listening for voice commands.

Project Structure

• agent.py: Main agent implementation and entrypoint
• mcp_client/: Package for MCP server integration
  • server.py: MCP server connection handlers
  • agent_tools.py: Integration of MCP tools with LiveKit agents
  • util.py: Utility functions for MCP client

Acknowledgements

• LiveKit for the underlying real-time communication infrastructure
• OpenAI for GPT-4o and text-to-speech
• Deepgram for speech-to-text
• Silero for Voice Activity Detection