The TNZ Model Context Protocol (MCP) client acts as a bridge between Claude Desktop and TNZ's messaging suite, allowing Claude Desktop to "act" rather than just "chat."
By integrating the TNZ MCP client with Claude Desktop, you empower your AI environment to interact directly with the TNZ CPaaS platform.
You'll be able to use natural language prompts to trigger SMS, Email and Voice messages, effectively giving your AI the ability to reach out to the real world.
Prerequisites
Before configuring Claude Desktop, ensure you have completed the following:
-
TNZ Account: You must have an active TNZ account with an API Auth Token.
-
Claude Desktop Installed: Ensure Claude Desktop is installed and running.
-
TNZ MCP Installed: You must have the tnz-mcp Python package installed. Refer to this guide: TNZ MCP Client
Understanding the Integration
Claude Desktop is designed to be extensible:
When you type a command like "Send an SMS to +6421...", Claude analyses the intent, routes it through the MCP configuration, and the TNZ Python plugin executes the message delivery.
Configuration Steps
To enable the integration, you must register the tnz-mcp server in Claude Desktop. The CLI supports two connection methods: Stdio (Standard Input/Output) and HTTP.
1. Locate or Create the Configuration File
Claude Desktop uses a JSON configuration file to manage external server connections. The location of this file depends on your operating system.
Possible locations:
You may be able to create this file manually, version dependent.
2. Edit the Configuration
Open the JSON file in your preferred text editor and add the mcpServers block using one of the methods below.
Stdio Mode (Recommended for Local Use)
This method allows Claude to spawn the TNZ MCP process directly. It is generally the most stable configuration for desktop use.
- Open your
claude_desktop_config.json file.
- Add the following JSON structure. If you already have other MCP servers configured, simply add "tnz-mcp" to the existing list.
Ensure tnz-mcp is installed and accessible in your PATH.
HTTP Mode (For Remote/Networked Servers)
Use this method if you are running the TNZ MCP server separately (e.g. via Docker or on a remote machine) and exposing it via a URL.
- Ensure your server is running in the terminal and confirm the host and port values:
tnz-mcp --transport streamable-http --host localhost --port 8000
- Open your
claude_desktop_config.json file.
- Add the following JSON structure, adjusting the URL to match your host and port. If you already have other MCP servers configured, simply add "tnz-mcp" to the existing list.
Verifying the Installation
Once you have saved your settings.json file:
Step 1: Restart Claude
Completely quit the Claude Desktop application (ensure it is not running in the background tray) and reopen it.
Step 2: Check for the Icon
Look for a "plug" icon or an integrations indicator in the Claude interface. You should see tnz-mcp listed as an active tool.
Step 3: Test Functionality
Interact with Claude using natural language to test the connection. Try the following prompt:
If configured correctly, Claude will confirm the action and send the SMS.
Troubleshooting
-
PATH Environment: If using Stdio, Claude must be able to find the tnz-mcp executable. If it fails, try replacing "command": "tnz-mcp" with the full absolute path to your Python executable (e.g., "/usr/local/bin/python3") and add args to call the module.
-
JSON Syntax: Ensure your config.json file is valid JSON. A missing comma or extra bracket will prevent Claude from loading any tools.
-
Server Running: If using HTTP mode, ensure the terminal window running the tnz-mcp server remains open.
-
Logs: Consult the official Anthropic documentation for instructions on how to access Claude Desktop error logs for detailed debugging information.