MCP Local Servers

An MCP local server is a program that runs on your machine and implements Model Context Protocol (opens new window)(MCP). It enables AI models to access and interact with local tools and data sources like files, databases, and APIs by running the server on the same machine as the AI model.

FEATURE AVAILABILITY

MCP is only available for the US datacenter.

Your MCP server acts as a regular API platform user and supports all API platform authentication methods. This lets you securely connect an MCP client to any API collection using standard credentials without remote setup. All Workato API collections are compatible with this feature.

MCP Local Server requirements

You must have the following components to use an MCP local server:

• Node.js v18.0.0 or later
• npm v8.0.0 or later
• An LLM client, such as Claude Desktop, Cursor, or Windsurf

Install a local MCP server

Complete the following steps to install a local MCP server:

1

Open your preferred MCP client. For example, Claude, Cursor, or Windsurf.

2

Locate your MCP configuration settings in the client, and update the config to use the Workato MCP server with the authentication token in the MCP URL or with the authentication token as a separate environment variable:

1

Replace the REMOTE_MCP_URL value with the remote MCP URL generated for your API collection.

Example configuration:

{
  "mcpServers": {
    "workato-apim": {
      "command": "npx",
      "args": ["@workato/mcp-server"],
      "env": {
        "TYPE": "apim",
        "REMOTE_MCP_URL": "https://mcp.workato.com/your-username/your-collection-id?wkt_token={YOUR_API_TOKEN}"
      }
    }
  }
}

1

Replace the REMOTE_MCP_URL and AUTH_TOKEN values with the remote MCP URL generated for your API collection and your authentication token.

Example configuration:

{
  "mcpServers": {
    "workato-apim": {
      "command": "npx",
      "args": ["@workato/mcp-server"],
      "env": {
        "TYPE": "apim",
        "REMOTE_MCP_URL": "https://mcp.workato.com/your-username/your-collection-id",
        "AUTH_TOKEN": "YOUR_API_TOKEN"
      }
    }
  }
}

Retrieve your remote MCP URL

Your API recipe collections and AI gateway collections provide unique MCP URLs.

Complete the following steps to retrieve the MCP URL for your API endpoint collection:

1

Sign in to your Workato account.

2

Go to Platform > API platform > API collections.

3

Select the collection you plan to retrieve the MCP URL for. The Endpoints page displays by default.

4

Locate the Remote MCP URL section and click the Copy URL button.

Copy the remote MCP URL

5

Click the Copy URL button to copy the MCP URL and replace (YOUR_API_TOKEN) with your API key. Refer to API keys for more information.

Add a remote MCP server to an AI model

You can use your REMOTE_MCP_URL to configure your local server to work with your AI model, such as OpenAI or Anthropic.

Refer to the following OpenAI example steps for more information:

1

Sign in to Workato.

2

Go to Platform > API platform > API collections.

3

Select the collection and copy the Remote MCP URL.

4

Go to your OpenAI account.

5

Select Prompts in the sidebar.

6

Click Tools > Create > MCP Server > +Add new.

7

Paste your remote MCP URL into the URL field.

Configure your MCP Server

8

Add a descriptive label for your MCP server.

9

Ensure that Authentication is set to Access token/API key.

10

Add your API key to the key field.

11

Click Connect. The connection establishes and a list of available tools displays.

12

Use the Approval drop-down menu to select the approvals required for the tools listed.

MCP Tools and Approval

13

Click Add to complete the setup.