diff --git a/scripts/classifier_adapters.py b/scripts/classifier_adapters.py
index fc5f37a..3021ec4 100644
--- a/scripts/classifier_adapters.py
+++ b/scripts/classifier_adapters.py
@@ -20,7 +20,6 @@ __all__ = [
     "GLiClassAdapter",
     "RerankerAdapter",
     "FineTunedAdapter",
-    "EmbeddingKNNAdapter",
 ]
 
 LABELS: list[str] = [
@@ -158,7 +157,7 @@ DEFAULT_EXEMPLARS: dict[str, list[str]] = {
         "Subject: Culture Fit Assessment\n\nAs part of our process, we ask all candidates to complete a short assessment.",
         "Subject: Skills Assessment\n\nWe'd like you to complete our online coding assessment before proceeding.",
         "Subject: Personality Assessment\n\nPlease complete the following assessment as the next step in our process.",
-        "Subject: Interview Feedback Survey\n\nThank you for interviewing — your feedback helps us improve our process.",
+        "Subject: Pre-interview questionnaire\n\nBefore we schedule your interview, please complete this brief skills survey.",
     ],
     "neutral": [
         "Subject: Application Received\n\nWe have received your application and will be in touch.",
@@ -194,6 +193,7 @@ DEFAULT_EXEMPLARS: dict[str, list[str]] = {
     ],
 }
 
+
 class ClassifierAdapter(abc.ABC):
     """Abstract base for all email classifier adapters."""
 
diff --git a/tests/test_classifier_adapters.py b/tests/test_classifier_adapters.py
index 793f306..aaea9ba 100644
--- a/tests/test_classifier_adapters.py
+++ b/tests/test_classifier_adapters.py
@@ -301,7 +301,7 @@ def test_default_exemplars_covers_all_labels():
     from scripts.classifier_adapters import DEFAULT_EXEMPLARS, LABELS
     for label in LABELS:
         assert label in DEFAULT_EXEMPLARS, f"DEFAULT_EXEMPLARS missing label: {label}"
-        assert len(DEFAULT_EXEMPLARS[label]) >= 1, f"{label} has no exemplar texts"
+        assert len(DEFAULT_EXEMPLARS[label]) >= 4, f"{label} needs >= 4 exemplars for k=3 voting"
 
 
 def test_default_exemplars_sparse_labels_have_at_least_four():
@@ -311,3 +311,14 @@ def test_default_exemplars_sparse_labels_have_at_least_four():
         assert len(DEFAULT_EXEMPLARS[label]) >= 4, (
             f"{label} needs >= 4 exemplars for k=3 voting to work reliably"
         )
+
+def test_default_exemplars_strings_are_formatted_correctly():
+    from scripts.classifier_adapters import DEFAULT_EXEMPLARS
+    for label, texts in DEFAULT_EXEMPLARS.items():
+        for text in texts:
+            assert text.startswith("Subject: "), (
+                f"{label!r} exemplar missing 'Subject: ' prefix: {text[:50]!r}"
+            )
+            assert "\n\n" in text, (
+                f"{label!r} exemplar missing double-newline separator: {text[:50]!r}"
+            )