marcus-margarites/FastMcpServer

MCP Server

This is a FastMCP implementation of a MCP server.

For more information about the server itself, see https://gofastmcp.com/getting-started/quickstart.

Running Locally

For running the application locally, serving HTTP, use the following command:

fastmcp run server.py:mcp --transport http --port 8000

Then you can use the different batch and JSON files present in the directory to test using CURL. For example, you can use the for initiating a connection:

curl -i -sS -X POST http://localhost:8000/mcp -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" -d @init.json

---

More information about CURL on Windows can be obtained at https://ss64.com/nt/curl.html and https://curl.se/docs/tutorial.html.

---

It is easier if you use the batch files, though.

But for that you'll need to grab the response of the init request, locate the "mcp-session-id" header and use its value in the next requests, updating the subsequent batch files with it.

date: Tue, 04 Nov 2025 19:16:56 GMT
server: uvicorn
cache-control: no-cache, no-transform
connection: keep-alive
content-type: text/event-stream
mcp-session-id: fa0c606c4361464280bcfc80b3ca89c5
x-accel-buffering: no
Transfer-Encoding: chunked

event: message
data: {"jsonrpc":"2.0","id":1,"result":{"protocolVersion":"2025-06-18","capabilities":{"experimental":{},"prompts":{"listChanged":true},"resources":{"subscribe":false,"listChanged":true},"tools":{"listChanged":true}},"serverInfo":{"name":"mvfm's mcp server","version":"1.17.0"}}}

Use the batch files in the following order:

• init.cmd
  • Don't forget to update the next batches with the "mcp-session-id" value.
• initialized.cmd
• list_tools.cmd
• call_add.cmd

Or... You can use the Bruno test collection...

Publishing to FastMCP Cloud

For publishing your MCP Server to FastMCP Cloud, you can follow this tutorial: https://gofastmcp.com/deployment/fastmcp-cloud.

After that, adapt the batch files so you can test pointing there instead of the local server. Or, as I said before, use the Bruno collection instead.

Connecting to ChatGPT

For connecting to ChatGPT you'll need to host your MCP Server on the Cloud (it can be FastMCP Cloud, it can he Python Anywhere or any other service where you can deploy applications and have a public IP address).

After that, test it and make sure everything is OK. Then, go to your ChatGPT client and:

• Click on your account, on the bottom-left of the screen;
• Select the "Settings" option;
• On the left panel, click on the "Connections" option;
• If you don't see a "Create" button in the top-right corner of the window:
  • Go to the bottom of the screen and click on the "Advanced Options" option;
  • Enable the "Developer Mode";
  • Click on the "back" button to go back to the Connections configuration panel;
  • You should now see the "Create" button.
• Click on the "Create" button in the top-right corner of the window;
• The "New Connector" window will appear;
• Fill it with the information about your MCP Server:
  • Name: The name of your tool.
  • Description: An optional description for your tool.
  • MCP Server URL: The URL of your server as determined by your provider (for example "https://mvfm2136.fastmcp.app/mcp").
• Check the "I understand and I want to continue" checkbox;
• Then click on the "Create" button.