halfaipg/glados-mcp

GLaDOS MCP Server

A Model Context Protocol server that provides GLaDOS-style voice commentary for AI development workflows, along with professional Kokoro TTS voices.

Overview

This project combines:

• GLaDOS Voice: Authentic Portal-style audio synthesis with characteristic commentary
• Professional Alternatives: 26 Kokoro voices for serious development work
• MCP Integration: Seamless integration with existing Model Context Protocol tools

Demo Videos (Turn on sound)

https://github.com/user-attachments/assets/f5623dd5-ac9b-4adc-8641-104d634504c7

https://github.com/user-attachments/assets/542fb5c6-03ab-4f4d-9e73-0d7f29a71c4e

Installation

1. Download Models (Required)

python download_models.py

2. Install System Dependencies

Linux (Ubuntu/Debian):

sudo apt update
sudo apt install portaudio19-dev python3-pyaudio

Linux (Fedora/RHEL):

sudo dnf install portaudio-devel python3-pyaudio

macOS:

brew install portaudio

3. Install the Server

cd glados-mcp
python -m venv venv
source venv/bin/activate  # Windows users: venv\Scripts\activate
pip install -e .

Configuration

Add this to your MCP client configuration:

{
  "mcpServers": {
    "glados-mcp": {
      "command": "./glados-mcp/venv/bin/python",
      "args": ["-m", "tts.server"],
      "cwd": "./glados-mcp",
      "env": {
        "PYTHONPATH": "./glados-mcp",
        "MCP_DEBUG": "1",
        "MCP_LOG_LEVEL": "DEBUG"
      }
    }
  }
}

Note: Make sure the paths match your actual installation directory. If you installed in a different location, adjust the command and cwd paths accordingly.

Usage

The interface is straightforward:

Default GLaDOS Voice:

speak("Your code is... adequate. I suppose.")

Professional Kokoro Voices:

speak("Analysis complete.", voice="af_alloy")
speak("Task finished.", voice="bm_daniel")

Available Voices:

list_voices()  # Returns all 26 Kokoro voices plus GLaDOS

Voice Options

• GLaDOS (default): Characteristic Portal-style commentary
• Kokoro Female US: af_alloy, af_aoede, af_bella, af_jessica, af_kore, af_nicole, af_nova, af_river, af_sarah, af_sky
• Kokoro Female British: bf_alice, bf_emma, bf_isabella, bf_lily
• Kokoro Male US: am_adam, am_echo, am_eric, am_fenrir, am_liam, am_michael, am_onyx, am_puck
• Kokoro Male British: bm_daniel, bm_fable, bm_george, bm_lewis

Project Structure

├── download_models.py   # Downloads TTS models (run this first!)
├── mcp.json            # MCP client configuration
├── glados-mcp/         # Main server code
│   ├── tts/            # TTS system
│   ├── models/         # Voice models (downloaded by script)
│   └── README.md       # Detailed instructions

Requirements

• Python 3.10+
• Audio output
• Internet connection for model download
• Linux: PortAudio system library (portaudio19-dev)

License

Apache 2.0

"The cake is a lie, but the voice commentary is real."