Full transcript of “How to Test MCP Server using Postman | Postman New feature unlocked”. Click a timestamp to open the video at that moment. Need another video? Use the YouTube transcript generator.
0:00 Hello everyone, welcome back to my channel Prince Automation Destination. This is Prince. I hope you are doing well. So today we are going to talk about how to test a MCP server with the help of Postman. But before that let us try to understand what is the need of this particular feature in Postman. So if you haven't heard the term MCP, MCP stands for model context protocol. It provides context to LLMs with the help of which LLMs can talk to data sources and can pull the data from data sources and can publish the results onto the tools. Right? So this is the purpose of MCP. We will soon try to understand right now as there are like many servers many MCP servers coming into the market
0:46 every day. So it becomes important to test them as well. So that is the reason Postman has introduced this particular feature. So we will be discussing about the same right. So without further delay let me jump onto the discussion. So we'll be validating MCP server with the help of postman. So what we'll be learning as part of current discussion we will try to understand what is MCP what is selenium MCP how to make a request in Postman how to validate the functionality of MCP server. connecting to Selenium MCP server from Postman and performing Selenium operations on Chrome browser with Postman and as a result we will be validating the functionality of Selenium MCP server. Now let me go to the official documentation of Selenium MCP right. So this is Selenium MCP uh
1:34 not Selenium MCP MCP I mean right. So what is MCP? MCP is an open protocol that standardizes how applications provide context to LLMs. Right? So LLM is large language model that is a AI model. So if we want to take some output from LLM we need to provide the context right. For example, we need to type in some information like act as a quality assurance engineer and create the test cases for this particular module. Right? So this is how we provide context and as a result it gives output. So let me show you from one example. Right? Suppose we have a functionality available for application which has a login page which has two fields username and password and login page has following options login button forget
2:19 password and remember me. So what we can do is we can type in to a AI tool such as chat GPT claude Google Germany act as a quality assurance engineer create manual test cases for the following functionality right so we are going to copy it and we are going to provide it as an input or as a prompt to the LLM right and as a result LLM will produce the output meaning the AI tool will produce the output right so now how this information came right so this information came from stakeholders which would have been docu documented somewhere in SRS or let us say it is documented in confluence right so instead of copying this particular context from this particular confluence and providing it to LLM what we can do is we can create or establish a connection between LLM and confluence
3:07 and we can read this particular information from the conference itself instead of copying and pasting so as a result we can get the context of this particular conference page right so this is what is the purpose of MCP So MCP provides the context to the LLMs right. So it is like a C port. So as like we have a C port using which we can connect to different peripherials. Similarly this MCP is a unified protocol using which we can connect to different data sources. These data sources could be confluence Jira. It could be GitHub. Right now let us try to understand why MCP. So it helps us building agents and complex workflows on top of LMS. Right? So by default LLM does not have capability to talk to data source such
3:52 as if it it want to talk to confluence it does not have that capability right. So with MCP it becomes possible and there are like many servers available in the market. So if I go to the official GitHub repository. So this is the official GitHub repository where we can take a look at different MCP servers, right? So if you want to for example talk to Git, if you want to talk to our local file system, GitLab, Google Drive, Google Maps. So what we can do is we can directly utilize these MCP servers in LLMs and we can talk to these uh data sources. Right? Now let us come back. Right? So LLMs frequently need to integrate with data and tools. Right? If LLM want to get the context of conflence, it need to be integrated, right? And this is where MCP comes into the picture. MCP provide
4:38 the growing list of pre-built integrations. So these are the pre-built integration already available. So instead of writing a custom code to connect to a particular data source, we can utilize these MCP servers, right? It has flexibility to switch between LLM. So we don't need to rely on a particular LLM. Instead, we can choose or switch between let us say GPT4. We can use claudet right we can use alarmama right and best best practices for securing your data right so when we'll be connecting to this mcp server what will happen is with this particular mcp we'll be running one mcp server that will be running locally right so this is the architecture right let us try to understand first of all we have mcp client right host with mcp client so this mcp client uh could be a cloud
5:23 right which has the capability to talk to mcp server right so this is the client client server server based architecture. So this is MCP server. Now let us say this particular MCP client right which is having a LLM want to talk to a data source. Let us say it want to talk to file system of your system. Meaning if for example we want to get the directory of D drive. So by default it cannot talk to this and suppose there is a uh like MCP server as I just showed there is a file system uh MCP server available right. So what we can do is we can utilize this particular MCP server which has capability to connect to this particular uh file system and then we can create a connection between this particular MCP client and this particular MCP server and as a result LLM would be able to talk to this particular data source and if you will
6:10 type uh in LLM that please fetch Directory and fetch all the directory of D drive it would be able to give us result and now suppose we want to for example access the database So what we can have is let us say we want to access Postquase database. So we have a Post case u Postgress SQL MCP server is also available right. So what we'll simply do is we will simply integrate Postquase MCP server um into our MCP client and as a result we will be able to access this particular postquest database right and now we can these are the local data sources. Similarly if you want to talk to outer world which is available on the internet so that is also possible. So we can have MCP server right and we can talk to the external APIs as well right.
6:56 So this is what is MCP server right? What is MCP? So it is a client server architecture where we have a client which want to talk to data sources and that is possible with the help of uh MCP server right now let us try to understand what is selenm because we are going to connect to selenium mcp from postman and we'll be validating the functionality of selenium mcp with postman right. So similar to uh these MCP server there is one MCP server introduced by Angie Jones right she is the developer of MCP server what she has done is she has provided the selenium functionality into this particular selenium MCP server so she has provided functionality of start browser navigate to a URL find element click element and
7:42 so on also she has provided the configuration as well how to connect to this particular MCP server right so this This is what is the configuration right now. Let me come back here and this is how we are going to do. So this is a user who want to validate selenium MCP server with the help of postman. So he will be creating a MCP request and then he'll be making a connection with selenium MCP server and he'll be trying to perform operations and as a result he will be validating selenium MCP server with the help of this particular postman. Right? So this is uh the UI right where we can check uh the MCP which is the new feature right and this is after we click on MCP this is how it look like right now let us go to postman
8:30 right and I'm simply going to click on new and here we will be seeing MCP so what we are going to do so we will see this particular request so if you will see there is two type of request stdio this is for local MCP server so most of the servers are going to be local in case we want to connect to remote MCP server we can choose as HTTP. So first I will try to connect a local MCP server and then I will try to connect a remote MCP server right. So now uh what I need to do I need to go to that cell in MCP server right uh and I need to copy this configuration and in postman I need to simply provide this right. So as soon as I provide this so this command says npx npx stand for running the node module hyphen y for yes
9:17 so it will download this particular mcp uh selenium node module and it will run it as a result selenium mcp server would be running locally. Now let me try to create a connection with this right. So it will create a connection with this. It may take some time. So let us wait for the same. So it is connected. So as soon as it is connected we are able to see these sections right. So we can see the tools right. So whatever the tools or capability defined by this particular MCP server. So we are able to see these options right. So in order to launch an application uh what we need to do we need to first launch a browser. So for that purpose she has provided this functionality start a browser. So for now she has provided support for Chrome and Firefox. So now let us test this functionality test this particular start
10:02 browser functionality. So I'm simply going to click on run and as a result it should open a chrome browser right. So she it has opened chrome browser. Now what we can do is we can navigate to a URL. Right? So let us say we want to navigate to source demo. Right? So I'll simply copy it and I'll click on this navigate option right and I'll provide this particular URL and as soon as I click on run I'll be able to see this particular uh URL is launched right user is navigated to this particular URL. Now let us try to enter the data and username. So for username this is the ID. So what I'm going to do so for entering the data what we need? We need send keys right? So here uh there are multiple options. First of all, we need
10:48 to choose the locator strategy. So I have taken ID, right? So I'm going to choose ID. And then we can enter the data. So that username is this one. So it is provided here itself. Right? So we can click on run. As a result, it should have entered standard user. Similarly, we can locate password as well. And instead of username, this time we will provide password. And we will copy the password as well. Right. So if I click on run, password should also be entered. Now let us locate login right. So for login the id is login button. Now I'll go for another functionality that is click element and here I'll be providing this
11:33 particular login button. Right? So for login we don't need to provide any text. We simply need to provide the locator. So if I click on run it should be able to login. So if you notice we are able to login. So this is how we can test the functionality of a selenium MCP server with the help of postman right. So this is one thing right now we so far we uh like tried to connect MCP server locally similarly we can connect to a remote MCP server as well for that purpose uh we can there is the MCP server like there could be multiple remote MCP server available. So one such server is provided by MCP which is the eco server just for demonstration purpose. So I'm going to connect it right and here we can type hello prince automation
12:20 destination right. So if I click on run it will give me the output right in the e code right. Similarly we can connect to other uh like other mcp servers. So for example we want to talk to uh let us say uh github right. So I'll be going to GitHub right and here somewhere there would be some configuration. So these are the functionality provided. We can create or update a file. We can push files. We can search repository. We can create a repository. Right? And if I come down we'll be able to see the configurations right. So if we want to go by docker approach we can use this. If you want to use npx that is not module based we can copy this. Right? So uh in selenium mcp
13:07 guest we didn't need uh the github credentials but in this case as we are going to talk to a github repository right we need to provide the github token right so I'm going to simply go here going to provide this particular request right and if you will notice it has environment so that's why environment is also populated and here we need to provide the token right and let me copy this particular token right so I have already kept in kept it I mean you can create your access token. Right? So here if you will provide let me click on connect and as a result it would be able to connect to my GitHub repository. Right? And it would be able to perform operations on my GitHub right. So let us say we want to create a repository. But before that let me
13:53 navigate to my uh like uh this particular GitHub repository. If I refresh it so I had a sample already created right. So there are like 26 repositories. Now let us do one thing. Let me uh provide the repository name. So we are going to create the repository. So let us say postman demo. Right? Whether we want to keep it private or uh public. Let us make it private. Whether we want to initialize readme or not, right? Let me click on run. Right? And as a result, it should be able to create the uh repository. So let me refresh and we should start seeing 27 repository. So postman demo is created. So this is how we can establish the connection between the data source
14:38 and the llm right. So this is but with the help of postman we are just validating this functionality right. Okay. So this is how we can validate the functionality. Apart from it what we can do is we can track the timeline as well. So whatever the request we are making we can track those whatever the operations performed right. So we can validate everything as like we do in case of APIs. We can store them as a collection. So suppose I want to give it name selenium mcp right or we can provide the uh collection name. Let us say we want to keep it in some new collection right. So we can simply go and store here right. So similar to whatever we do for rest APIs we can store these are collection these are the collection. We can export it. We can save it. uh we can share it
15:25 with our team members. So this is how we can validate any selenium mc any MCP server admin with the help of postman and this is what I wanted to cover as part of current discussion. Thank you for watching and I would request you to please like, share and subscribe. Thank you once again.
