2025-02-13 –, RoboCon
When maintaining UI tests, a lot of time is wasted on fixing broken locators or rerunning tests with synchronization problems. Could those simple tasks maybe be done by someone else? And could those fixes be applied in real time during the execution? Let's do some experiments with Large language models and see they can support us.
I will use the new RF 7.1 listener enhancements to implement a solution for self healing tests during runtime. RF will send broken locators and other errors to a LLM and apply the fix directly during runtime. The test execution will proceed normally and the changes are collected in a small report (with line number of the robot framework file where the changes were applied).
It is just a proof of concept with the potential to help engineers reducing the time they need to spend on rerunning and fixing broken tests.
In person
Categorize / Tags:AI LLM SELF-HEALING
Lessons Learned:How to use the new listener API changes with RF 7.1.
How to deploy your own LLM simply.
Prompting LLM for self healing.
Being sceptical about LLMs
Intermediate RF User, Advanced RF User
Father and husband, musician, movie nerd and Robot Framework enthusiast.
Providing test automation expertise and guidance as a QA Engineer at DB Schenker.
Board member of Robot Framework Foundation.