Generating Automated Tests with MCP + LLM
-
40 min
The talk explores the capabilities of the MCP protocol. In particular, I will cover what MCP is, why it is needed, and which problems it solves. I will demonstrate how even the weakest locally deployed model can sometimes outperform proprietary or very powerful open-source LLMs in certain responses. During the talk, we will walk through an example implementation of an MCP client and MCP server, and have a closer look at how to automate browser testing using MCP Playwright on a fully locally hosted infrastructure.
This talk will be valuable for QA engineers and developers seeking to improve the reliability and efficiency of testing with the help of LLMs.