Prompt Regression
What it is
Captured-prompt replay. (input, baseline_output) corpus rerun against current model + prompt; diff against baseline.
When to use it
Catching silent prompt drift across model upgrades or prompt edits.
Example gates
Future slice — needs captured baseline corpus.
See also
- The pyramid — where this mode sits relative to the others
- Writing a gate — how to scaffold a new gate in this mode
- Decision tree — "I just wrote feature X — what tests do I owe?"