Working with MCP servers

With Voximplant, you can connect to MCP (Model Context Protocol) servers via WebSockets and act like an MCP client.

The Model Context Protocol is an open standard that enables AI models to communicate with external data sources and tools through a unified, secure interface. By acting as an MCP client, Voximplant can bridge its advanced communication features—such as real-time speech recognition (ASR), high-quality synthesis (TTS), and intelligent turn detection—directly with MCP servers that host private databases or third-party APIs. This allows you to build sophisticated AI agents that can listen via Voximplant's VAD, retrieve real-time context from an MCP resource, and then act by triggering an MCP tool to update a CRM or send an email, all within a single automated conversation.

Learn more about MCP on the official website.

To create an MCP client:

1. Require the Modules.MCP module in your VoxEngine scenario.
2. Create an MCP.Client instance (via the MCP.createClient method) and provide your parameters.
3. Call the mcpClient.listTools method to get the available tool list from an MCP server.
4. Call the mcpClient.callTool method to call the specified tool and receive an answer from the MCP server
5. Listen to the MCP.Client events (via the MCP.ServerEvents event list) and implement you application business logic.

Here is the complete scenario example on how to create an MCP client with Voximplant: