eval_config.yaml

# Planning-mode Creative Director: research + storyboard, no production claims.
metrics_to_run:
  - multi_turn_task_success
  - no_fake_asset_urls

custom_metrics:
  - name: no_fake_asset_urls
    custom_function: |
      def evaluate(instance: dict) -> dict:
          response = instance.get("response") or {}
          text = "".join(
              p.get("text", "") for p in (response.get("parts") or []) if p.get("text")
          )
          lowered = text.lower()
          banned = (
              "storage.googleapis.com" in lowered
              or "mock_cut" in lowered
              or "c2pa" in lowered and "signed" in lowered
          )
          return {
              "score": 0.0 if banned else 1.0,
              "explanation": (
                  "Planning replies must not invent GCS URLs or claim C2PA signing."
                  if banned
                  else "No invented asset URLs or signing claims."
              ),
          }