Playwright + LLM + MCP: Using AI for API and UI Testing

Learning objective

This training shows how to connect a Model Context Protocol (MCP) server for Playwright with AI tooling (e.g., GitHub Copilot / Cursor) and modern LLM approaches such as tool calling (function calling) and Retrieval-Augmented Generation (RAG). The goal is to accelerate exploratory testing and the development of automated tests for web applications.

By the end of the course, participants will be able to:

• Explain what tool calling (function calling) and RAG are, and how they apply to software testing.
• Describe the role of MCP and its architecture in the Playwright ecosystem.
• Run LLM-assisted exploratory testing sessions.
• Generate and maintain Playwright tests (UI and API) with AI assistance.
• Assess when AI support improves productivity and when it increases risk.

Agenda

1) Modern LLM techniques

• LLM fundamentals (how tools like ChatGPT operate)
• Tool calling / tools: connecting the model to external capabilities
• Retrieval-Augmented Generation (RAG): improving answer quality using internal documentation
• AI agents: concepts, theory, and practical applications

2) MCP and Playwright

• MCP server architecture
• Installation, configuration, and integration with Playwright (JS/TS)
• API overview: accessibility snapshots, navigation, and element actions

3) MCP in hands-on testing

• Prompt-driven exploratory testing
• Generating test cases from user stories

4) Test automation with AI

• Using Copilot / Cursor for test automation
• Connecting Copilot / Cursor with MCP: a "prompt → code → run → fix" workflow
• Practical workshops:
  • Automating REST API checks with Playwright
  • Automating UI tests with MCP + Playwright
• Productivity review: what to write manually vs what to generate

5) Risks, ethics, and maintainability

• Model hallucinations vs test stability and quality
• Versioning prompts and ensuring reproducibility
• Cost analysis and return on investment
• MCP and Playwright roadmap: what to watch next

Prerequisites

Who is this for?

The course is intended for testers who have basic programming skills in JavaScript/TypeScript and understand Playwright fundamentals. Familiarity with Git is recommended.

What to prepare

• A laptop with Node.js (LTS), Git, and VS Code
• An AI coding tool configured (e.g., GitHub Copilot or Cursor)

A detailed setup checklist is provided after registration.

Teaching methods

Workshops (~65%) plus lecture-style explanation (~35%). Each practical block ends with a short retrospective and knowledge sharing. Participants leave with working code and a prompt history.

Training materials

• Slide deck (PDF)
• Git repository with branches for each module
• Prompt notebook (Markdown)
• A curated board with links and theory

Benefits

• Practical skill: controlling a browser via MCP, increasingly in demand.
• Faster automation development with AI support.
• Stronger exploratory testing with AI as an assistant.
• Better risk management: checklists and guidance on when not to trust LLM outputs.
• Knowledge sharing with peers: prompts, templates, and repository patterns.