Agent Yeager

Agent Yeager is a premium, mobile-first intelligence terminal web application for voice communication and AI assistants. It features a stunning, GPU-accelerated tactical HUD interface and real-time WebRTC audio streaming.

• Tactical UI/UX: Dark matte theme, 8-point grid, layered glassmorphism shadows, and an animated environmental background.
• Spotify Voice Control (MCP Integration): Seamlessly control music playback, search tracks, and manage your Spotify queue using natural voice commands processed through a local AI assistant and routed to the Spotify Model Context Protocol (MCP) server.
• Push-to-Talk (PTT): Hero interaction button featuring an animated CSS oscilloscope waveform.
• Real-Time Audio: Secure WebRTC streaming between devices.
• Terminal Logs: Authentic rolling terminal interface for system status and command override.
• Progressive Web App (PWA): Installs natively on mobile devices.

• Node.js (v14 or higher)
• npm

1. Clone the repository:

   git clone <your-repository-url>
   cd webrtc-audio-bridge

2. Install dependencies:

   npm install

1. Start the server and Electron app:

   npm start

2. The server will run on https://0.0.0.0:3000 (HTTPS is required for WebRTC microphone access).
3. Open https://localhost:3000 on your desktop or https://<YOUR_LOCAL_IP>:3000 on your mobile device. You will need to accept the self-signed certificate warning in your browser.

• server.js: Node.js Express server handling Socket.io signaling and static file serving.
• mobile.html: The mobile-first tactical terminal interface.
• manifest.json & sw.js: PWA configuration.
• beatles_bg.jpg: The background environment image.

The tactical HUD relies on CSS variables defined in mobile.html. You can customize the look by adjusting the :root variables:

• --accent: Primary interaction color (default: #FF003C)
• --success: Online/Connected indicator (default: #22C55E)
• --text-main: Primary text color (default: #EAEAEA)

Future development of Agent Yeager includes fine-tuning the underlying local AI models using Domain-Specific Language (DSL) training. By defining a strict grammar and vocabulary tailored specifically to tactical operations and Spotify control directives, we plan to train the model to yield highly deterministic, ultra-low latency command execution. This will eliminate hallucinated responses and vastly improve offline reliability.

if spotify mcp does not connect manually run

node auth.js

MIT License