Questions I am trying to answer by building things.

Small investigations around software testing, AI, automation and human judgment. Most of them are unfinished. Each one starts from a question and records what I tried, what I noticed and what did not work, as it happens.

  1. Can an AI agent investigate a failed E2E test?

    When an E2E test fails, can an agent work out why, and say how sure it is?

    An example of the format. Not started yet.