MCP meta tool

The NeoLoad Web Model Context Protocol (MCP) server offers a meta tool endpoint. It's a lightweight alternative to the standard MCP endpoint.

The standard endpoint loads all 45 tool definitions into your AI assistant's context window. The meta tool exposes only two: mcp_discover and mcp_execute. Your AI assistant discovers actions on demand and runs them through a single interface, so more of the context window stays available for your prompts and responses.

This topic covers both tools and endpoint configuration.

Before you start

To connect to the meta tool endpoint, you need a working MCP connection to NeoLoad Web and a compatible AI assistant that supports the MCP.

Available tools

Use mcp_discover to browse the action catalog and mcp_execute to run any action by name. Choose a tool based on its format and examples:

Returns a list of all available actions with their names, descriptions, and parameter schemas. Your AI assistant calls this tool at the start of a session or when it needs to find the right action for your request.

Each entry in the response contains the following fields:

Field

Description

name

The unique name of the action.

description

A summary of what the action does.

parameters

The JSON schema for the action's input parameters.

Example response:

Copy 
[
  {
    "name": "list-workspaces",
    "description": "Get a list of all workspaces you can access.",
    "parameters": { ... }
  },
  {
    "name": "get-result",
    "description": "Get detailed information about a test result.",
    "parameters": { "resultId": { "type": "string" } }
  }
]

Runs a specific action by name with parameters. Your AI assistant calls this after it identifies the right action from the catalog.

The request includes the following fields:

Field

Description

action

The action name from the catalog.

parameters

A JSON object with the action-specific input parameters.

Example request:

Copy 
{
  "action": "list-workspace-results",
  "parameters": {
    "workspaceId": "abc-123",
    "pageSize": 5,
    "sort": "-startDate"
  }
}

The response contains the action name and the result data:

Copy 
{
  "action": "list-workspace-results",
  "data": {
    "results": [ ... ]
  }
}

Connect to the meta tool endpoint

To use the meta tool, point your AI assistant to the meta tool URL instead of the standard endpoint URL:

To connect Cursor, follow these steps:

  1. Open your Cursor settings, then open the Tools & Integrations tab.

  2. Select New MCP Server.

  3. Add the meta tool server configuration with the URL for your region:

    • US Region: https://neoload-v2.saas.neotys.com/mcp-server/mcp-v2

    • EU Region: https://neoload-v2-eu.saas.neotys.com/mcp-server/mcp-v2

    Copy 
    {
      "mcpServers": {
        "nlweb-meta-mcp": {
          "command": "npx",
          "args": [
            "mcp-remote",
            "https://neoload-v2.saas.neotys.com/mcp-server/mcp-v2",
            "-y",
            "--header",
            "Authorization:${AUTH_HEADER}"
          ],
          "env": {
            "AUTH_HEADER": "<your NLW API token>"
          }
        }
      }
    }

  4. Enter your API token in the authentication settings.

To connect Claude Desktop, follow these steps:

  1. Open your Claude Desktop configuration file.

  2. Add the meta tool server configuration with the URL for your region:

    • US Region: https://neoload-v2.saas.neotys.com/mcp-server/mcp-v2

    • EU Region: https://neoload-v2-eu.saas.neotys.com/mcp-server/mcp-v2

    Copy 
    {
      "mcpServers": {
        "nlweb-meta-mcp": {
          "command": "npx",
          "args": [
            "mcp-remote",
            "https://neoload-v2.saas.neotys.com/mcp-server/mcp-v2",
            "-y",
            "--header",
            "Authorization:${AUTH_HEADER}"
          ],
          "env": {
            "AUTH_HEADER": "<your NLW API token>"
          }
        }
      }
    }

  3. Replace <your NLW API token> with your actual API token.

You can use both endpoints side by side. They share the same API token and access the same data.

What's next

Here's where to go after you've set up the meta tool endpoint: