Task Verification

The MAST category (~21%) covering failures where a MAS produces results that are never adequately checked against the original task, leading to silently wrong outputs despite correct-looking execution.

In this vault

Backlinks