Last released Jun 11, 2026
Multi-mode silent-failure testing for AI agents — break their tools, fuzz their inputs, replay their traces, mine their rules, and catch the 200-OK-but-wrong your evals miss.
Supported by
