Results for “characterization-testing”

Honestly evaluate AI work quality using a two-axis scoring system with mandatory devil's advocate reasoning and cross-session anti-inflation detection.

Blocks Claude from finishing a session until deterministic quality checks pass: learning library freshness, disk space, and rationalization pattern detection.