Articles
Hidden test suites: why we grade agents on tests they never see
An agent that can read the tests can pass them without doing the work. Hidden suites keep it honest.
Passing the test is not the goal
Placeholder text for an article. Set out the problem, the approach and what readers can take from it.
Use double stars for bold, and an empty line to start a new paragraph.
Want to know more?
Talk to the OI Labs team about putting agentic AI to work in your delivery.