Agent

Build and deploy real-time voice and video AI agents with LLM, STT, and TTS pipelines.

Updated May 27, 2026

Search Tags:#ai #llm #agents #video #voice #realtime #telephony

Ready to run

npx skills add https://github.com/rex009x/mush-talk --skill agent-rex009x

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: Agent
Source: https://github.com/rex009x/mush-talk/tree/main/.agents/skills/agent
Command: npx skills add https://github.com/rex009x/mush-talk --skill agent-rex009x

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires vision-agents, uv.

What problem does it solve?

This Skill solves the complexity of building real-time, low-latency conversational AI by providing a unified framework for managing audio/video streams, LLM pipelines, and call lifecycles.

Core Features & Use Cases

Real-time Voice/Video Pipelines: Seamlessly integrate STT, TTS, and LLM providers for interactive AI agents.
Call Lifecycle Management: Automatically handle call routing, turn-taking, and session management for voice and video calls.
Production Deployment: Scale agents using HTTP server modes, Docker, and Kubernetes with built-in telemetry and monitoring.

Quick Start

Use the Agent skill to initialize a new project and start the local development server for your voice agent.

Frequently Asked Questions about Agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build and deploy real-time voice and video AI agents?▼

To build real-time voice and video AI agents, you orchestrate LLM, STT, and TTS pipelines using a unified framework that manages audio/video streams and call lifecycles. This requires configuring agent lifecycle methods, transport layers, and provider plugins for low-latency interaction.

How does call lifecycle management work for conversational AI?▼

Call lifecycle management for conversational AI automatically handles call routing, turn-taking, and session management for voice and video calls. This ensures seamless interaction by managing the state and flow of real-time audio streams alongside LLM processing.

What is the best way to scale real-time voice agents for production?▼

The best way to scale real-time voice agents for production is using HTTP server modes, Docker, and Kubernetes. This deployment approach includes built-in telemetry and monitoring to ensure low-latency interaction and horizontal scaling across instances.

Do I need vision-agents and uv to develop real-time conversational interfaces?▼

Yes, you need vision-agents and uv configured in your environment to develop real-time conversational interfaces. These dependencies support the framework's ability to manage video streams and handle project environment setup for the AI pipelines.

Can I integrate telephony systems with LLM pipelines for AI agents?▼

Yes, you can integrate telephony systems with LLM pipelines for AI agents. The framework supports telephony integration alongside STT and TTS providers, allowing you to route external calls directly into your conversational AI workflows.