Test Your AI Tools With Postman
Video Transcript
In this crucial lesson, you'll learn how to connect your Agent Payment Model Content Provider (MCP) server directly to Postman. This process is essential for testing your tools and validating their responses before you start building complex agent prompts.
🧪 Agent Payment + Postman Integration Tutorial
Speaker: Stephanie Goodman
Topic: Connecting MCP Server to Postman and Testing Agent Tools
[00:05]
Hi, I'm Stephanie Goodman with Agent Payment.
In this video, I'll show you how to:
– Connect your MCP server to Postman
– Hook it up to Agent Payment
– Test the available tools with real requests
[00:18]
I’m doing this on a Linux machine, so the file locations might be slightly different if you’re using Windows.
When using an MCP server in Postman:
– You’ll need to create a new collection (Postman won’t let you add a server to an existing one)
– Locate the Agent Payment MCP server file
– If you downloaded it via our NPM command (from the previous video), this is where you’ll find it
Make sure Studio is selected in the STDAO corner—it should be the default.
[00:57]
Next, let’s add our environmental variables:
- Budget key and API key → You’ll find these in your Agent Payment dashboard
- Haven’t signed up yet? Watch our first video for the setup steps
Paste your budget key and API key into Postman.
– The API key is in the Account Details tab
– Add a metadata instance ID (optional, but helpful for tracking requests in the dashboard)
[01:57]
Once configured, you’ll see the available tools in Postman.
These are the tools you've added to your budget.
You can view:
– What parameters each tool requires
– How they respond by submitting test requests
Just input a test function, hit Run, and observe the response below.
[02:43]
This is a great way to test tools before connecting your agent.
You can confirm:
– Required parameters
– Expected response structure
– Data format your agent will receive
This helps refine your agent prompts, knowing exactly how tools will behave.
[03:09]
In the dashboard, you’ll also see:
– The request you just sent
– What was passed to the tool
– How the tool responded
These logs are visible whether:
– You send the request via Postman/MCP server
– Or your LLM agent sends it directly
This is extremely helpful for debugging or designing your prompt flows.
[03:20]
You can also check the tool page to see required parameters—
they should match what you see in Postman.
That’s how you test tools before hooking up your agent.