promptdojo_

Golden evals earn autonomy — every macro proves itself per intent — step 5 of 7

The promotion rule, in one expression: a macro is eval-passed only if EVERY golden case passed. One failure — any failure — and it stays review-required. There is no "passed most of them" tier.

results maps case id to True/False. Fill in the condition.

Expected output:

where-is-my-order: eval-passed
reset-password: review-required