WebRTC MCP Chat Server

┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ Web Users │ │ MCP/Cursor Users│ │ Background │ │ │ │ │ │ Agents/CLI │ ├─────────────────┤ ├─────────────────┤ ├─────────────────┤ │ • Video/Audio │ │ • Text Chat │ │ • API Calls │ │ • Screen Share │ │ • MCP Tools │ │ • CLI Commands │ │ • WebRTC P2P │ │ • Resources │ │ • HTTP/REST │ └─────────┬───────┘ └─────────┬───────┘ └─────────┬───────┘ │ │ │ └──────────────────────┼──────────────────────┘ │ ┌─────────────────┐ │ Chat Server │ │ (Remote) │ ├─────────────────┤ │ • Socket.IO │ │ • HTTP/REST API │ │ • WebRTC Signal │ │ • Temp Rooms │ │ • Auto Cleanup │ └─────────────────┘

The included CLI tool provides everything background agents need:

# Create secure temporary room chat-room create --expires 120 --created-by agent-1 # Join room with credentials chat-room join <roomId> <token> <username> # Send secure messages chat-room send <roomId> <token> <username> "Hello world" # Get room information chat-room info <roomId> <token> # Check server health chat-room health # Interactive mode for agents chat-room interactive

# Agent 1: Create room ROOM_INFO=$(chat-room create --expires 60 --output json) ROOM_ID=$(echo $ROOM_INFO | jq -r '.roomId') ROOM_TOKEN=$(echo $ROOM_INFO | jq -r '.roomToken') # Agent 2: Join and communicate chat-room join $ROOM_ID $ROOM_TOKEN agent-2 chat-room send $ROOM_ID $ROOM_TOKEN agent-2 "Task complete"

#!/bin/bash # In your pipeline ROOM_INFO=$(chat-room create --expires 30 --created-by ci-pipeline --output json) echo "Deployment room: $(echo $ROOM_INFO | jq -r '.joinUrl')" chat-room send $(echo $ROOM_INFO | jq -r '.roomId') \ $(echo $ROOM_INFO | jq -r '.roomToken') \ ci-bot "Deployment started for ${COMMIT_SHA}"

// Node.js service const response = await fetch(${CHAT_SERVER}/api/create-temp-room, { method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify({ expiresInMinutes: 30, createdBy: 'service-a' }) }); const { roomId, roomToken } = await response.json(); // Share credentials with other services

- Room Creation: ~10ms
- Message Delivery: ~5ms
- Health Checks: ~2ms
- Memory per Room: ~1KB
- Concurrent Users: 1000+ per instance

- Stateless Design- No database required
- Memory-Only Storage- Fast and efficient
- Auto-Cleanup- Expired rooms automatically removed
- Load Balancer Ready- Health check endpoint included
- Multi-Instance- Use sticky sessions for WebSockets

{ "status": "healthy", "serverUrl": "https://your-app.com", "remoteMode": true, "activeRooms": 5, "temporaryRooms": 3, "connectedUsers": 12 }

# Railway (Recommended) railway up # Vercel vercel --prod # Test deployment export CHAT_SERVER_URL=https://your-deployed-app.com chat-room health # Test ngrok header functionality npm run test:ngrok-headers

We welcome contributions! Areas of interest:

- Additional deployment platforms
- Enhanced security features
- Performance optimizations
- More CLI commands
- Integration examples

MIT License - see LICENSE file for details.

Ready to deploy secure, temporary chat rooms for your background agents?

Then:chat-room create --expires 60 --created-by my-agent

🚀Your agents can now communicate securely!

An AI voice toolkit with TTS, voice cloning, and video translation, now available as an MCP server for smarter agent integration.

Connect your AI agents to Google-Meet, Zoom & Microsoft Teams through tl;dv

Upfirst is an AI phone receptionist for small businesses. Review call transcripts, then fix the greeting, knowledge, and transfer rules from your AI client.

A server for text-to-speech generation using the AivisSpeech engine.

Provides audio input and output capabilities for AI assistants.

A Node.js MCP server for the Japanese text-to-speech software Bouyomi-chan.

Enables communication between multiple AI characters with simultaneous voice playback using VLC.

Generates text-to-speech audio with automatic playback using the Chatterbox TTS model.