
faultgate 0.3.0: a refund over the limit fails, even inside the noise
Business rules on tool arguments, an honest INCONCLUSIVE verdict with sample size math, vendor trace import, and zero cloud telemetry. Release gates for agents.
Field notes and empirical failure analysis tagged with #release-gate. (2 articles)

Business rules on tool arguments, an honest INCONCLUSIVE verdict with sample size math, vendor trace import, and zero cloud telemetry. Release gates for agents.

You run the same model five times on the same thirty inputs and pass rates swing from 0.60 to 0.87. A fixed threshold is a coin flip. faultgate measures your spread instead.