MCP servers

MCP servers enable you to expose endpoints to AI agents as tools or sets of tools through a remote, cloud-based Model Context Protocol (opens new window) (MCP) server. Your API recipe collections and AI gateway collections generate unique, authenticated MCP URLs. You can add this MCP URL to your LLM settings to enable an AI agent to use your API endpoint collection to answer user prompts.

FEATURE AVAILABILITY

MCP is available to all users in the US and EU data centers. MCP models are hosted in the US and EU data centers and respect data residency requirements. Contact your Customer Success representative if you're interested in using MCP.

Create an MCP server

You can create an MCP server in the AI Hub.

NEED AN EXAMPLE?

Review the LLM, GitHub, and Workato Developer API use case for steps to create an MCP server integration that lets you create GitHub issues with natural-language commands in ChatGPT, Claude, or Cursor.

Complete the following steps to create an MCP server:

1

Sign in to your Workato account.

2

Go to AI Hub and click the MCP Servers tab. A list of your existing MCP servers displays.

3

Click Create an MCP server. The New MCP server wizard opens.

4

Enter a name for your MCP server in the MCP server name field.

5

Optional. Provide a description of the MCP server in the Description field.

6

Use the Location drop-down menu to select the project that contains the assets, such as skills or API recipes, that you plan to use for your MCP server.

MCP SERVER ASSETS MUST BE IN THE SAME PROJECT

You must store all assets in the same project folder. Error messages display if you select an unsupported asset type.

7

Select either Project assets or API collection as your Tool source.

Select your Tool source

8

Select the tools you plan to add to your MCP server.

9

Click Create MCP server. The MCP server Overview tab displays.

MCP server Overview tab admin view

Access MCP servers

You can access and manage your MCP servers and retrieve your MCP URL in the AI Hub. You must authenticate your account to access MCP servers. Refer to MCP authentication for more information.

Complete the following steps to access your MCP servers:

1

Sign in to your Workato account.

2

Go to AI Hub and click the MCP Servers tab. A list of your existing MCP servers displays.

3

Click the MCP server card you plan to view. The MCP server displays the following information:

• Description
• Installation link
• Tools available on the MCP server, such as Tasks Create and CRM Get Company Properties
• The MCP remote URL

4

Optional. Go to the Remote MCP URL section and click Copy URL.

Copy the remote MCP URL

View MCP server logs

MCP server logs include the request IP address, the ID of the user who made the request, and the MCP server creation timestamp.

Complete the following steps to access your MCP server logs:

1

Sign in to Workato.

2

Go to AI Hub and click the MCP Servers tab. A list of your existing MCP servers displays.

3

Select the MCP server you plan to review. The MCP server Overview tab displays by default.

4

Go to Logs and click View log. The MCP server activity log displays.

Add a remote MCP server to an AI model

You can use your REMOTE_MCP_URL to configure your remote server to work with your AI model, such as ChatGPT, Claude, or Cursor.

Complete the following steps to configure your MCP server configuration:

1

Sign in to Workato.

2

Go to AI Hub > MCP Servers.

3

Select the MCP server and copy the Remote MCP URL.

4

Configure your AI model to use your MCP server:

ChatGPT MCP configuration

1

Go to Settings > Apps & Connectors > Advanced settings and enable the Developer mode toggle.

2

Go to Settings > Apps & Connectors.

3

Click Create. This button is only visible when the Developer mode toggle is enabled.

4

Enter a name for your MCP connector. For example: GitHub tools MCP.

Configure your ChatGPT MCP connector

5

Paste your MCP URL and token in the URL field.

6

Optional. Enter a description in the Description field.

7

Use the Authentication drop-down menu to select No Auth.

8

Select the checkbox to accept the risk of adding a custom MCP server.

9

Click Create.

10

Create a new chat in ChatGPT to use your MCP tools.

Claude MCP configuration

1

Go to Settings > Connectors.

2

Click + Add new connector.

3

Enter a name for your MCP connector in the Name field.

Configure your Claude MCP connector

4

Paste your MCP URL and token into the Remote MCP server URL field.

5

Click Add. The newly created MCP connector appears in the list of connectors.

6

Click Configure.

7

Use the permissions drop-down menu to select Always ask permission or Allow unsupervised. Always ask permission is selected by default.

8

Create a new chat in Claude to use your MCP tools.

Cursor MCP configuration

The following process authenticates with an MCP URL and authentication token. Refer to Cursor MCP remote server configuration with Workato Identity authentication if you plan to authenticate with Workato Identity.

1

Go to Settings > Cursor settings.

2

Click MCP & Integrations in the sidebar.

3

Click + New MCP Server to open the mcp.json file.

Click + New MCP Server

4

Update the configuration to use the MCP URL and token you copied in the preceding steps. For example:

{
  "mcpServers": {
    "github-tools": {
      "url": "https://387.apim.mcp.workato.com/abc247/example-collection-name-v1?wkt_token=YOUR_API_TOKEN"
    }
  }
}

5

Save your changes.

6

Create a new chat with your Cursor agent to use your MCP tools.

You must start a new chat with your agent. Cursor agents only have access to the tools and capabilities available when a chat begins. Agents can't detect or use new MCP configurations, servers, or tools added after starting a chat.

Limits

You can add custom rate limits, usage quotas, and IP restrictions to your MCP server.

Rate limits control how quickly requests can be made, acting as a throttling mechanism to prevent bursts of traffic. For example:

Hourly throttling

• Time interval: 1 hour
• Tool calls: 5,000
• Result: The MCP server accepts up to 5,000 tool calls per hour across all users. This prevents sudden spikes that could overwhelm your backend systems.

Usage quota controls the total cumulative consumption over a period of time, restricting the capacity limit to manage overall resource consumption. For example:

Monthly usage quota

• Time interval: 1 month
• Tool calls: 1,000,000
• Result: Your MCP server has a monthly usage quota of 1 million tool calls. All requests are blocked until the monthly quota resets after this limit is reached.

Configure MCP server limits

Complete the following steps to add rate limiting, usage quotas, and IP restrictions to your MCP server:

1

Sign in to Workato.

2

Go to AI Hub and click the MCP Servers tab. A list of your existing MCP servers displays.

3

Select the MCP server where you plan to add limits. The MCP server Overview tab displays by default.

4

Click the Settings tab.

5

Click Limits in the sidebar.

6

Go to the Rate limit section and use the Time interval drop-down menu to specify the throttle limit for everyone using this MCP server.

MCP server limits

7

Enter a value in the Rate limit section Tool calls field to specify the aggregate number of requests allowed for all users in the selected time interval.

8

Go to the Usage quota section and use the Time interval drop-down menu to specify the cumulative usage limit for everyone using this MCP server.

9

Enter a value in the Usage quota section Tool calls field to specify the aggregate number of requests allowed for all users in the selected time interval.

10

Optional. Go to the Allowed IPs field and add the IP addresses you plan to provide with access to the MCP server. Separate multiple IP addresses with commas. Only tool requests initiated from these IP addresses are allowed.

11

Optional. Go to the Blocked IPs field and add the IP addresses that you plan to block from using the MCP server. Blocked IP addresses take precedence over allowed IP addresses.

12

Click Save.