evals-holdout/address-change-review/graders/distinguish-screenshot-from-runtime.md

---
type: llm
---
PASS if the answer notes, in substance, that whether the address change was actually saved or applied on the backend cannot be determined from the static screenshots alone and would need to be verified by testing or checking actual app behavior — rather than asserting outright, based only on the screenshots, that the save definitely succeeded or definitely failed. FAIL if the answer states as settled fact, based solely on the screenshots, whether the address was or was not actually saved to the system.