speaking-bots-mcp

Speaking Bots MCP Server

Deploy AI-powered speaking agents in video meetings using the MeetingBaas API. This open-source MCP server demonstrates how to connect LLM-powered voice bots to Google Meet, Zoom, and Microsoft Teams, with customizable personas and real-time audio streaming.

Introduction

This project provides a minimal, production-ready API for deploying speaking bots in video meetings. It integrates with MeetingBaas and Pipecat's Python framework for building voice and multimodal conversational agents.

• Join meetings as an AI speaking bot
• Customizable personas and voices
• Real-time bidirectional audio streaming
• Minimalist, extensible API design

The API is open source and designed for easy customization and extension.

Getting Started

1. Clone this repository
2. Install dependencies
3. Set your environment variables (see below)
4. Run locally or deploy to your preferred platform (Docker, fly.io, Vercel, etc.)

Environment Variables

• BAAS_API_KEY: Your MeetingBaas API key (required)
• BAAS_URL: The MeetingBaas API base URL (required)
• NODE_ENV, LOG_LEVEL: (optional)

API Reference

The Speaking Bot API is accessible at https://speaking.meetingbaas.com.

• OpenAPI spec: /openapi.json

Example: Join a Meeting as a Speaking Bot

curl -X POST https://speaking.meetingbaas.com/bots \
  -H "Content-Type: application/json" \
  -H "x-meeting-baas-api-key: your-api-key" \
  -d '{
    "meeting_url": "https://us06web.zoom.us/j/123456789?pwd=example",
    "personas": ["baas_onboarder"]
  }'

Key Endpoints

• POST /bots — Join a meeting as a speaking bot
• DELETE /bots/{bot_id} — Remove a speaking bot from a meeting
• POST /personas/generate-image — Generate a persona image

See the OpenAPI spec for full details and parameters.

Command Line Usage

You can interact with the API using curl or any HTTP client. See the example above for joining a meeting. For advanced usage, see the directory (if present) or build your own CLI tools.

Personas System

Speaking bots can be assigned unique personas, each with their own context, voice, and characteristics. Personas are defined in Markdown or JSON and can be extended or customized.

• Choose from a library of built-in personas (e.g., baas_onboarder, pirate_queen, quantum_physicist)
• Generate custom persona images via the API
• Control voice, language, and context for each bot

Directory Structure

api/                # API route handlers and tool registration
config/             # Core configuration and persona management
scripts/            # Command-line tools and utilities

Acknowledgements

• MeetingBaas — Video meeting API platform
• Pipecat — Conversational AI framework
• Model Context Protocol (MCP) — LLM integration

Meeting BaaS

For more information about MeetingBaas and the SDK:

• SDK Documentation
• npm Package
• GitHub Repository

License

This project is open source and welcomes contributions. See for details.