Learning Objectives

After completing this lesson, you'll be able to:

• Describe how FME acts as an MCP client with the MCPCaller.
• Configure the MCPCaller to connect to an MCP server. 
• List available tools from an MCP server with the MCPCaller. 
• Call a specific MCP tool and receive a response. 

In this lesson, you will:

• Optional: Watch a demonstration video (if you have not attended the live training).
• Scroll down to read the activity instructions below and follow the steps in your lab.
• Complete the quiz at the end of this lesson.
• Click 'Next' to mark the lesson complete.

Resources

• FME and MCP Strigo training lab with Weather MCP server
  • If you're taking a live, online training course, you will have access to the lab within the live event. Please do not request an on-demand lab unless you are working on the course materials outside of the live, online course time. 
  • If you're taking the course on-demand, please request an MCP and FME on-demand lab to work through the exercises. 
• Complete workspace | C:\FMEData\Workspaces\MCPandFME\DynamicWeatherMCPCaller-start.fmw

MCPCaller

The MCPCaller transformer connects to an existing MCP server to work with MCP tools, resources, and prompts. The MCPCaller is the key component that enables FME to act as an MCP client, calling MCP tools on a server and receiving the output. 

Within your FME workspace, the MCPCaller connects to an MCP server and issues a request. If the MCP server requires authentication, the MCPCaller uses a web connection to authorize access to the server. Each MCP server has a unique URL that you enter into the MCPCaller to connect to it. Using the URL, the MCPCaller connects and requests the MCP server when it runs. If the MCP server requires authentication, you may need to enable Override Server and enter the MCP server URL to override the URL you saved in the web connection. Every MCP server URL must support streamable HTTP transport. 

The MCPCaller can perform various types of requests to the MCP server, from inquiries to executions. You can list and interact with tools, resources, and prompts. Instead of building a custom connector for each external service, FME can connect to an MCP server, inspect the tools it exposes, and call them consistently. You can then parse, transform, write, route, or combine the result with other data using standard FME transformers.

Depending on the MCP request method, the MCPCaller will require additional input data along with the request to the MCP server. For example, the call tool method requires you to select the tool and provide additional input as JSON.  

The MCPCaller output varies by method, although all communication between the MCP server and client uses JSON as the protocol standard. To return the raw JSON of the MCP response for that operation, enable the Include JSON Response option in the Advanced section. This may be useful for inspecting the full server response or accessing fields the transformer doesn't otherwise expose as attributes. 

The MCPCaller is an optional input transformer, meaning you don't need an input to run it. When you provide input records to the MCPCaller, it runs once per record received to the Input port. If you don't connect any input to the transformer, it will run only once. For more information on optional input transformers, see Transformers with an Optional Input Port. 

List Tools

The List Tools request returns a list of all available MCP tools from the server. It is a discovery inquiry about the MCP server you are working with and is the best place to start when working with MCP. Running a List Tools request confirms that the MCPCaller can connect to the server and returns the currently available tool names, descriptions, and schemas.

List Tools requires no additional input beyond setting the MCP server to query; however, the Advanced options let you control whether the MCPCaller returns each tool as an individual record or combines them into a single list. 

The output from the MCPCaller List Tools includes the tool name, title, description, and JSON schemas for input and output. You may send the List Tools results to an AI Connector transformer or local AI model to analyze which tool best performs the task you need and which inputs you require to run the tool. 

Call Tool

The Call Tool method requests the MCP server to run the tool you specify. If the MCPCaller has multiple inputs, it will make multiple Call Tool requests, one for each input record. If you don't provide input to the MCPCaller, it makes the Call Tool request only once and outputs a single record. 

Once you select the tool to call, the MCPCaller parameters update to show you the input JSON schema the tool expects, along with a Tool Input section to configure. The JSON input to a tool may come from an Attribute or Text, or from a File containing the input JSON. To customize the tool input from your data, reference attribute values in the JSON Text. 

The output from a Call Tool request creates a _data attribute containing the tool call's results in JSON. You can further parse the JSON into attributes with the JSONFragmenter to continue working with the MCP output in your workspace. 

For more information on working with JSON in FME, check out these resources:

• Tutorial: Getting Started with JSON
• Transforming JSON using the JSONExtractor, JSONFlattener, and JSONFragmenter

To see a complete MCPCaller workflow integration with ArcGIS, check out our TGIF: Thank Goodness It's FME video on Bringing ArcGIS Tools into FME with the MCPCaller.

Exercise

To start working with MCP and FME, Frank will connect to an existing weather MCP server using the MCPCaller in FME Workbench. He'll use the MCPCaller to discover what tools his MCP server provides using List Tools, then use another MCPCaller to manually call a tool to return weather data and information. Frank won't connect these MCPCallers, as he's just exploring each call individually. In the next lesson, he'll build on these with an AI model that chooses the right tool based on user input.

In this exercise, you will:

• Connect to an MCP server and list its available tools with the MCPCaller transformer.
• Use another MCPCaller to call an MCP tool and return weather information.  

1) Open FME Workbench

• Start FME Workbench (2026.2 or newer). 
• Open a blank workspace. 

2) Add an MCPCaller to List Tools

Frank already has an external MCP server running and ready to connect to. It provides weather information based on an input location, such as forecasts and air quality metrics. Configure the MCPCaller to connect to this server. 

• Add an MCPCaller to the canvas and open its parameters. 
• Leave the Credential Source set to None. 
• Set the MCP Server URL to http://weather-mcp:3000/mcp.
• For the MCP Method, select List Tools. 

• Click OK. 
  • Optionally, add a bookmark around the MCPCaller and title it List Tools. This helps you track what each MCPCaller transformer does in your workspace. 
• Run the MCPCaller transformer. 

FME connects to the MCP server, discovers its available tools, and returns a feature for each available tool.

•  Inspect the data cache for the Output port. 
  • There should be 17 available tools, and FME displays each tool's information as attributes. Aligning with MCP protocol, each tool provides its name, description, input schema, and expected output schema. 

3) Call a Tool

Now that you've used List Tools to see the available tools on the weather MCP server, you will use another MCPCaller to call a specific tool. The two MCPCallers are stand-alone for now; in the next exercise, you'll modify them to chain a dynamic tool call with AI. 

• Add another MCPCaller to the canvas and open its parameters. 
  • Do not connect this MCPCaller to the first one. The MCPCallers are standalone for now. 
• Again, leave the Credential Source set to None and set the MCP Server URL to http://weather-mcp:3000/mcp.
• Set the MCP Method to Call Tool. 
  • The Input Schema and Tool Input sections will appear. The Input Schema displays the tool's expected input as JSON, while Tool Input is where you provide the JSON to send with the tool request. 
• For the Tool Name, click the ellipses and select weather_forecast.
  • The Input Schema will update and show the schema for this specific tool. 

• Scroll down to the Tool Input section. For JSON Text, use the drop-down arrow to select Open JSON Editor...
• Replace the current text in the editor with the following JSON. 

{
	"current_weather": true,
	"forecast_days": 7,
	"latitude": 49.2827,
	"longitude": -123.1207,
	"precipitation_unit": "mm",
	"temperature_unit": "celsius",
	"timezone": "America/Vancouver",
	"wind_speed_unit": "kmh"
}

Copying and Pasting in Strigo Labs

This course requires a lot of copying and pasting to avoid typing out JSON and descriptions. To copy and paste content into your Strigo lab:

• Copy the content from the FME Academy.
• Paste the content into the Strigo lab's clipboard. 
• In the lab's virtual machine, paste (Ctrl + V) the content in the desired location.  

If you're having trouble pasting into the Strigo lab, log into the FME Academy in a browser in the lab and copy (Ctrl + C) and paste (Ctrl + V) directly within the lab. 

• Expand the Advanced section and set Include Response JSON to Yes. 

• Click OK to close the MCPCaller parameters. 
  • Optionally, add another bookmark around this MCPCaller and title it Call Tool to keep track of what this MCPCaller does. 
• Run this MCPCaller.
  • FME calls the tool and receives the response from the MCP server. The response comes as formatted JSON. 

• Inspect the Output and view the _json_response attribute. 
  • When viewing the _json_response, set Syntax Highlighting to JSON. 

4) Save Workspace

• Save your in-progress workspace.
  • You can save your workspace to C://FMEData/Workspaces/MCPandFME. 
  • You will continue with this workspace in the next exercise. 

So far, you've successfully connected to the weather MCP server, viewed its available tools, and called a tool directly with the MCPCaller. In the next lesson, you'll build on this by using an OpenAIConnector to dynamically select and call MCP tools based on user input, rather than configuring each tool call manually.