MCP integration

This topic describes how to run Controller Service processes through external AI agents, using the Model Context Protocol (MCP).

Note: This integration is available from version 26.3 as a private Beta release.

The integration requires installation of the Controller Service package. For details, see Run the Controller Service from the CLI.

About MCP

The Model Context Protocol (MCP) is an open standard that enables AI agents to securely connect with external tools and data sources. When you make a natural language request in the AI assistant chat, the MCP client translates this into a standardized protocol format and sends it to the Controller Service MCP server.

The MCP server then communicates with the Controller Service to execute the appropriate operations, and returns the response. This seamless flow transforms your conversational requests into actual operations, eliminating the need to manually navigate the UI to run and manage load tests, load generators, and virtual users.

The MCP server should work with any AI tool that supports MCP, but has been specifically tested on Microsoft Visual Studio Code with GitHub Copilot, and on Claude Desktop.

Implement the MCP integration

The MCP server is included in your Controller Service package. The following process enables you to install the MCP server and use its tools in chat.

To install the Controller Service MCP server:

  1. Copy the MCP executable, McpControllerService, from the Controller Service package to the local machine.

  2. Open the AI client and access the executable file. Run it to install the Controller Service MCP server as an extension for the AI client.

    The Controller Service MCP server folder includes an mcp.json file, containing default communication parameters for tools such as VS Code.

    For tools that need a different format or name (for example, claude_desktop_config.json), configure the file manually. The file name and location depend on the tool.

    Example:

    Copy code 
    {
        "mcpServers": {
            "McpControllerService": { 
                "type": "http", 
                "url": "http://127.0.0.1:3001/mcp" 
            }
        }
    }

  3. Find the installed Controller Service MCP server in the client Extensions list, and start it.

Run Controller Service processes

The MCP server supports various Controller Service tools, including running a scenario, stopping a test immediately, and opening the test run results in Analysis. You can retrieve the full list of supported tools by making a request in the AI chat.

Type your required task into the AI assistant chat box. For example, you could type in the chat "Run a test with 50 Vusers for 10 minutes on LG 11.222.333.44 for this script: <path to script>".

Note:  

  • Later, or more advanced, AI agent model versions are likely to provide faster and more stable results.

  • Some requests in the chat may require that you click allow before proceeding, due to security settings.

See also