Skip to main content

Video · 4:36

Build an AI QA Agent that reads Jira, runs Playwright, and closes the ticket

In this 4.5-minute demo, an AI agent built in Cursor takes a fixed Jira bug, picks the most relevant tests from Wopee.io, runs them with Playwright, and posts the test report back to Jira before moving the ticket to Done. The agent combines three MCP servers: the Wopee.io MCP server, Playwright and Atlassian. The video also shows how the same approach can turn specifications from Notion, Confluence or SharePoint into tests that run in Wopee.io.

Book a demoSet up the Wopee.io MCP server

Chapters

  1. 0:00The basic flow: a prompt and a URL to test results
  2. 0:49The Wopee.io MCP server, plus Playwright and Atlassian
  3. 1:44Demo: read the Jira bug and move it to In Progress
  4. 2:01Pick the most relevant tests from Wopee.io
  5. 2:30Playwright runs the tests
  6. 2:40Test report on the Jira ticket, ticket moved to Done
  7. 3:05Same approach: specs from Notion, Confluence or SharePoint
  8. 3:46Recap: an agentic colleague in your workflow
  9. 4:18Built with Cursor and three MCP servers

Transcript

Cleaned from the video's captions.

0:00The basic flow: a prompt and a URL to test results

Hi there. I would like to show you some more features Wopee.io can offer. I was doing multiple demos in the last months, and I found out that some people keep asking about additional use cases. So it's time to show you something more. This is the typical use case I often demo. It's just three basic steps: you provide your prompt and URL. Then we collect all the data about the app that our agent could collect during the crawling, based on the prompt, and then we provide test results. We have the test cases, we run them, and we can give you the results. But this is just a fraction of what Wopee.io can offer. So how it works: as I said, a prompt, then you have tests and reports, and then you have a battery of tests that you can execute later.

0:49The Wopee.io MCP server, plus Playwright and Atlassian

Apart from that, you can also use our Wopee.io MCP, and this opens a completely new galaxy for you in test automation and in boosting efficiency in your web app testing. Let me quickly explain how it works. Wopee.io MCP has multiple tools which you can easily integrate into your IDE. You can use Cursor, ChatGPT, Claude Code, whatever MCP client you have. Apart from this, for my demo I decided to bring two more tools. One is Playwright, with all the tools it offers, and the other is Atlassian, where you can fetch data like a specification from Confluence or a bug report from Jira. I combined these tools to build this demo, to show you how it could work.

1:44Demo: read the Jira bug and move it to In Progress

Imagine you have a bug and it's already fixed. So now I'm running my agent, combining Playwright, Atlassian and Wopee.io. First I am fetching the details about the ticket from Jira, and then I'm moving the ticket to In Progress.

2:01Pick the most relevant tests from Wopee.io

Once this is done, I am fetching all the test cases from Wopee.io to find out which ones could be most relevant. I am selecting the top three of these tests to execute them. Now I'm running the MCP tools to fetch all the data about the test cases, and I'm already prioritizing and preselecting the top five. Or not me, but the agent I built. Once I have it, I'm posting the information about the selected tests to Jira.

2:30Playwright runs the tests

I'm testing the ordering feature, and you can see that Playwright is now running the tests and the products are reordered.

2:40Test report on the Jira ticket, ticket moved to Done

Once this is done, I'm collecting all the details and preparing a test report, and this test report is added to Jira as a comment. Once this is done, I am moving the ticket to Done, and then my agent is finished. This is one of the use cases where you can use MCP, run it with a simple prompt and integrate all the tooling together.

3:05Same approach: specs from Notion, Confluence or SharePoint

This opens a completely new world, where you can use the same approach to fetch the specification from Notion, Confluence or SharePoint, use it to generate higher-quality tests, and then execute them in Wopee.io. The same can be done for new functionality. If you already have tests, you can easily push them into Wopee.io, so your colleagues can maintain them there, and then run the other agent to execute the tests you prepared. Then you can combine all these things together and automate more easily.

3:46Recap: an agentic colleague in your workflow

So what was this demo about? As I said, I fetched the bug, then I fetched the tests and prepared the test plan. Basically, I preselected the top five tests, executed three of them, and kept updating Jira while it was running. This is an agentic colleague, and you can integrate it into your workflow. Your developers can now ask this agentic tester to rerun the bug fixes.

4:18Built with Cursor and three MCP servers

This demo was run with Cursor, and as I said, I integrated Atlassian, Wopee.io and Playwright as MCP servers to show you how it could work. But there is much more. If you would like to try it, just go to Wopee.io, where you can find more information and details.

Try it yourself