Intermediate~45 minHazelJS
Evals and Reliability
Gate agent quality with @hazeljs/testing and @hazeljs/eval—golden tasks, forbidden tools, CI reporters.
- Authors
- editorial-team
- Published
- Last reviewed
Make evals a merge requirement: describeAgent suites, golden datasets, and reliability signals beyond vibes.
Learning outcomes
- Author expectTools / assertAgentResult cases
- Wire eval CI to block regressions