Where exactly are you using MCP Servers in your companies and how do you think, its going to improve your daily QA activity improvements ?

We are using MCP as a single integration layer across our QA services. For functional testing, it connects to Jira, TestRail, and databases to design and update cases. In test automation, it triggers Playwright/Pytest runs through CI/CD and pushes logs back to tracking tools.

For load & performance, we are planning to orchestrate JMeter/K6 executions and collects Grafana/NewRelic metrics.

Also, in security testing, we are in process to run OWASP ZAP and SAST scans, auto-logging vulnerabilities into Jira. This way, all our QA workflows run through a consistent MCP protocol with governance and auditability built in.

FOr automating my application based on use cases

Answer Karthik KK’s Question for a chance to win a ShiftSync Giftbox
 

 

I haven’t had the chance to use MCP Servers in my current work yet, but from what I understand, they could be really valuable in QA activities. MCP Servers provide a standardized way to connect tools, resources, and data pipelines, which means instead of manually switching between systems, QA teams can have a more unified workflow.

We are planning to use MCP Servers primarily for test automation result analysis by integrating them with Jira and Confluence. The idea is to enable failure comparison between test results and the associated documentation or tickets in these tools, allowing us to better understand the root cause of failures.

I am Thinking of using MCP server from the first stage of SDLC to the last stage.

1)Requirement Analysis/User Story analysis->Draft questions->Email them to the Reporter

2)Test Case Writing in a format which we use and upload it to Jira(BDD or Simple Test Case format)

3)Generating Realistic test data to test (Currently using MCP server here)

3)Smoke Testing->Functional Testing->Basic Non functional Testing

4)Test Reporting for all the project

5)Will be creating seperate agent for each step(Multi agent)

6)Defect reporting with the log file with log analysis (Currently using MCP server here)

7)Preparing the Demo /KT document of the userstories

8)Chatbot which will act as query resolver who will resolve your issue related to project.LIke how to run the payment flow,or how to shutdown the port

Using this in the form of client.
Integrating within VS Code for fetching the functions or keywords already created within the organization to reduce redundancy.
Also, this can improve the QA practices by not letting us invest more time as well as letting the new member to not spend a lot of time getting the idea of already created code, he/she can directly start coding automation and keep getting the pre-created keywords.

This is how we can get AI to help us design TC(s) with the help of MCP. 

As an individual who shifted from being a manual tester to a QA Engineer, I am up-skilling myself to learn about MCPs. In my current org, I am planning to integrate MCP into the process pipeline and make the testing lifecycle more streamlines and reduce the TAT.

By enabling MCPs , the manual efforts currently in place can be reduced significantly, and use the time saved to improve other aspects of our product.

MCPs & AI tools are a great tool for QA engineers , mainly in daily QA activities which are majorly repetitive on a daily basis.
 

I hope to implement MCPs and AI agents in my org by end of next 2 quarters.

I haven’t had the chance to use MCP Servers in my current work yet, but from what I understand, they could be really valuable in QA activities. MCP Servers provide a standardized way to connect tools, resources, and data pipelines, which means instead of manually switching between systems, QA teams can have a more unified workflow.

We haven't started using it in our company yet. However, I am aiming to give it a go to support my test data creation, code review and debugging. 

Answer Karthik KK’s Question for a chance to win a ShiftSync Giftbox
 

 

While I am not using MCP Servers yet, I see them as a strong enabler for AI-driven QA automation, helping to move from fragmented testing tasks towards a more unified, intelligent, and scalable process.

🎊Thank you everyone for participating in this challenge!  And special thank you to ​@executeautomation for hosting it.

🎉We’re happy to announce the winner of this challenge: ​@ndaine 

​@ndaine Please keep an eye on your email, we will reach out shortly to arrange your giftbox delivery, or A Udemy course of your choice, as well as a shareable certificate.

Stay tuned to ShiftSync for more events!👀

Thank you, Karthik, for this question. I have attended a couple of webinars from you over the past few weeks, and they have been insightful. Learnt a lot and looking forward to  ​@executeautomation ​@Mustafa ​@PolinaKr.

Where exactly are you using MCP Servers in your company, and how do you think they will improve your daily QA activities? - Made me to think not using advanced reasoning model though but using human mind. 

It all started a few months ago when I learned what MCP is, as it was newly released by Claude. I thought it was just another release among the many in the AI space.

I became aware of Playwright MCP thanks to #Debbie from Microsoft PW, who made me appreciate Playwright MCP more than the original MCP. I simply explored it for a few hours.

MCP stands for Model Context Protocol. The world has heard about MCP, and there are a couple of things I have explored.

I'm not going to write about Type C here, as you may have already read several articles on the topic. In my view, MCP is a standard through which our AI applications will receive the correct context from our systems.

For example, with Canva MCP, you can create designs just by sending a prompt, such as: "I'm planning to create a LinkedIn post on MCPs. Can you create a thumbnail?" or you can ask, "How many designs have I created now?"

Another example: If you use Tricentis Tosca, it has recently launched Tosca MCP, where you can run test cases or create playlists, all possible with a prompt.

For a beginner, just remember three components: MCP Host, MCP Server, and MCP Client.

The key participants in the MCP architecture are:

• MCP Host: The AI application that coordinates and manages one or multiple MCP clients.
• MCP Client: A component that maintains a connection to an MCP server and obtains context from an MCP server for the MCP host to use.
• MCP Server: A program that provides context to MCP clients.
• One more important thing - there are two transport methods: local/STDIO and http/remote servers which you need to be careful with. You can see the config of any MCP servers in the mcp.json file. If you send any data local servers like Playwright MCP process it in your machine while canva with remote config sends the data to canva ext server - which some cases against compliance policies. 

Practical Example: I will explain this using VS Code as the MCP Host, with GitHub Copilot and Playwright MCP Server.

Once we install the PW MCP server in VS Code and start the MCP Server, VS Code with Copilot acts as the host. When a user enters a query, Copilot checks the available servers with the help of an AI model like GPT-5 or Claude Sonnet 4 and asks the MCP server to use its specific tool (e.g., PW has a browser_navigate tool that can navigate to a URL). Every MCP server has tools like this. The PW MCP server can open a website, and the server sends this information to the host. Copilot can then assess it and show us what is happening. Without it, it couldn't navigate and get this context.

We will discuss more about PW MCP in later posts. For now, let's address the second query: How do you think it will improve your daily QA activities?

It's still GenAI, but think of it like going to a restaurant. We know the place delivers food, but the server (waiter) provides tools like the menu to help us make choices and get what we want.

Practical examples:

• You can install the Excel MCP server to retrieve test cases. If you're using Selenium, you can ask it to create boilerplate code.
• You can use Azure DevOps MCP to trigger pipelines. Once completed, you can use MCP to check logs, fix failures, make code changes, push the code to Git, and trigger the pipeline again.
• For a custom MCP server, you might need help from AI to create, use, and fix issues. You can use it for custom project work, such as using SQL Server MCP to create custom prompts to retrieve data from the database.
• Creating a exploratory Co-pilot tester based on PW MCP with custom instructions
• Using data from DB to prepare the proper project specific test data format - synthetic data 

It's not 100% accurate, but it works when you know what you're looking for. Based on further prompts, you can direct it better. For example, if Copilot has created a Selenium script and you have some functions in place, you can ask it to analyze the functions and use them in the script it created.

MCP is not going to replace anyone, but it makes some tasks easier—if and only if you know what you're doing and what Copilot is doing.

Where exactly are you using MCP Servers in your company? ADO MCP is the one we have started using, and the results look promising. It is still in the POC phase, as it is still AI and can hallucinate. We are the pilots driving them to make our work possible.

Image is created by OpenAI’s image model.

Thank you ​@PolinaKr  -  today I received the gift and happy to have it.