Agentic QA is making big promises, but traditional test automation is still faster, more specific and more accountable. Automated tests run quickly and repeatably, and inform both humans and agents how to interact with game code without breaking existing features. Writing test automation has always had a learning curve, but like most things, coding agents are helping make it more accessible.

In this workshop, videogame automation consultant Andrew Fray will take a real QA smoketest and turn it into automation: first by hand, using GameDriver’s HierarchyPath queries to robustly interact with the game, and then using GameDriver’s MCP server to let a coding agent finish the job. From there, he’ll demonstrate creating agent guidance that generates maintainable and trustworthy tests, and building a skill to search the commit history for bugfixes that deserve regression tests. Andrew will even cover data-driven, parametric setups that grow your test suite as the game matures.