Scalable Playwright Framework with AI

Because the way we build and maintain test automation frameworks is evolving fast, AI is becoming a practical part of this workflow.

We will begin presenting a brief overview of the AI & MCP (Model Context Protocol) concepts.

From there, participants will clone a GitHub repository containing a lightweight Playwright and TypeScript setup. Together, we will start adding layers to this project step-by-step.

My idea is firstly to start using the playwright codegen feature to record a simple test on an ecommerce website and understand the baseline structure. We will discuss custom-instructions files and include it in our project (Playwright specific) to guide the AI behaviour for our specific project context. I will point out the disclaimer regarding AI knowledge cutoff date (April 2024) and how to use context7 MCP config to make sure the agent uses up-to-date official documentation. Then, we will bring the GitHub Copilot Agents – starting with the Planner agent to scan the application in a browser using PW MCP server, generate a test plan and store it into a file.

Then we will use the Generator Agent to convert some of the test plan scenarios into executabile Playwright tests and provide support for generating Page Object Model. Finally, we will explore how the Healer Agent can help fix broken tests, significantly reducing maintenance time. Once the tests are stable, we will run them into a GitHub Actions CI/CD pipeline, demonstrating how smoothly the setup integrates into continuous testing workflows.

By the end of the workshop, participants will walk away with both conceptual understanding and a ready-to-extend project template that proves how AI can enhance test automation daily workflows.