diff --git a/results_analysis/tests/test_core.py b/results_analysis/tests/test_core.py
new file mode 100644
index 0000000..7cf986e
--- /dev/null
+++ b/results_analysis/tests/test_core.py
@@ -0,0 +1,62 @@
+import pytest
+from results_analysis import core
+
+
+def test_create_confusion_matrix_basic():
+    previous = [
+        {"run_id": "r1", "decision": "PASS"},
+        {"run_id": "r2", "decision": "FAIL"},
+        {"run_id": "r3", "decision": "PASS"},
+        {"run_id": "r4", "decision": "FAIL"},
+    ]
+    new = [
+        {"run_id": "r1", "decision": "PASS"},  # TP-like
+        {"run_id": "r2", "decision": "FAIL"},  # TN-like
+        {"run_id": "r3", "decision": "FAIL"},  # FN-like
+        {"run_id": "r4", "decision": "PASS"},  # FP-like
+    ]
+
+    matrix_dict = core.create_confusion_matrix(previous, new)
+    assert set(matrix_dict.keys()) == {"TP", "TN", "FP", "FN"}
+    assert matrix_dict["TP"] == 1
+    assert matrix_dict["TN"] == 1
+    assert matrix_dict["FP"] == 1
+    assert matrix_dict["FN"] == 1
+
+
+def test_confusion_matrix_class_to_json():
+    cm = core.ConfusionMatrix(TP=2, TN=3, FP=1, FN=0)
+    as_json = cm.to_json()
+    assert as_json == {"TP": 2, "TN": 3, "FP": 1, "FN": 0}
+
+
+def test_get_deltas_detects_changes():
+    previous = [
+        {"run_id": "a", "decision": "PASS"},
+        {"run_id": "b", "decision": "FAIL"},
+        {"run_id": "c", "decision": "WARN"},
+    ]
+    new = [
+        {"run_id": "a", "decision": "PASS"},
+        {"run_id": "b", "decision": "PASS"},
+        {"run_id": "c", "decision": "FAIL"},
+    ]
+    deltas = core.get_deltas(previous, new)
+    ids = {d["run_id"] for d in deltas}
+    assert ids == {"b", "c"}
+
+
+def test_empty_inputs_handling():
+    # Both empty lists should yield zero-matrix and no deltas
+    prev, new = [], []
+    matrix_dict = core.create_confusion_matrix(prev, new)
+    assert all(v == 0 for v in matrix_dict.values())
+    deltas = core.get_deltas(prev, new)
+    assert deltas == []
+
+
+def test_input_validation_typing():
+    with pytest.raises((TypeError, ValueError)):
+        core.create_confusion_matrix("notalist", [])
+    with pytest.raises((TypeError, ValueError)):
+        core.get_deltas([], "notalist")