If you are working with a web application, playwright + frontier LLM is incredibly capable. They added some recent features to make this sort of use case go a lot smoother:
If you set this up correctly, you can have a main agent issue natural language testing instructions to this playwright agent which returns a natural language summary of what it experiences. This is the sort of thing where I begin to get interested in the idea of agents working while I sleep.
Comments
If you are working with a web application, playwright + frontier LLM is incredibly capable. They added some recent features to make this sort of use case go a lot smoother:
https://playwright.dev/docs/release-notes#version-159
If you set this up correctly, you can have a main agent issue natural language testing instructions to this playwright agent which returns a natural language summary of what it experiences. This is the sort of thing where I begin to get interested in the idea of agents working while I sleep.