Module 5 — Connecting to the Outside World

Connecting to the Outside World

You built the app. Now you need to list it in the app store — and plug it into your other devices. This module is about portability: making your MCP server useful beyond the Postman tab it was born in.

What you'll be able to do
1

Export MCP server configs for Claude Desktop, Cursor, and VS Code and know exactly where to paste them

2

Add your server to Agent Mode — closing the loop between building a server and using it inside Postman

3

Understand the pathway to publishing on the Postman API Network and why verified trust signals matter more for MCP than for regular APIs

 

Export & Configure for MCP Hosts

Think of an MCP host like a smart home hub — a Philips Hue bridge, say. The bulbs (your tools) don't magically appear in the app just because they exist. You have to register them by telling the hub the right address. Claude Desktop, Cursor, VS Code with Copilot all work the same way: each one has a config file waiting for your server's command or URL. Once it's in there, the host discovers your tools automatically on startup.

Postman generates the exact config block you need. Pick your host below to see it — annotated so you know what every field actually means.

Interactive — select your MCP host
Choose a host
Claude Desktop
Cursor
VS Code + Copilot
 

Agent Mode — The Postman-Native Host

Here's the callback to the prerequisite course: Agent Mode is an MCP host. It speaks MCP natively. Which means the server you just built in the previous module can talk directly to your Postman agent — no config file editing, no path copying, no restarting anything.

The flow is four steps. Walk through them below to see exactly how it works.

Step-through — from MCP request to Agent Mode
 

Publishing on the Postman API Network

This isn't a requirement — it's an opportunity. The Postman MCP Catalog is where millions of developers discover servers to connect their agents to. If you built something genuinely useful, this is how other people's agents find it.

🏅
Why trust matters more for MCP than for regular APIs
With a regular REST API, a developer reads your docs and consciously decides whether to integrate. With an MCP server, an AI agent can call your tools autonomously — often without a human actively watching. The verified publisher badge isn't vanity. It's the signal that tells developers (and their agents) that they're connecting to the real, official, maintained thing — not an impostor server with a suspiciously similar name.

The checklist below covers everything you need to get your server into the catalog. Check items off as you go. All four required items must be done before you submit — but the fifth one is the step most people forget, and it's the one that actually drives discovery.

Publishing requirements — check off as you go
Road to the MCP Catalog
0 / 5
🎉 Ready to submit!
Email api-network@postman.com with a link to your public MCP collection. The team reviews it and, on approval, forks it into the official catalog.
 

See the Full Flow in Action

The video below walks through exporting a config for one MCP host, pasting it in, restarting the host, and sending a real request from outside Postman. Then a peek at what a well-published server looks like in the catalog.

 
Video Demo — Export to Host + API Network Walkthrough
Replace this block with the embedded lesson video when ready
 
🌎 Try This
 

Export your server's configuration for whichever MCP host you use most (or most want to try). Paste it in, restart the host, and send it one real request from outside Postman.

 

That moment when an external tool calls something you built in Postman — when the response comes back and it worked — that's the payoff. Everything else in this course was setup for that.

🏠
Course Complete
Back to Portfolio