diff --git a/rerun_analysis_tool/tests/test_core.py b/rerun_analysis_tool/tests/test_core.py
new file mode 100644
index 0000000..d3156a0
--- /dev/null
+++ b/rerun_analysis_tool/tests/test_core.py
@@ -0,0 +1,53 @@
+import pytest
+import pandas as pd
+from rerun_analysis_tool import core
+
+@pytest.fixture
+def sample_runs_data():
+    return [
+        {
+            'run_id': 'r1', 'status': 'WARN', 'unknown_rate': 0.2,
+            'rerun_helps': 1, 'rerun_shifts': 0, 'rerun_hurts': 0
+        },
+        {
+            'run_id': 'r2', 'status': 'PASS', 'unknown_rate': 0.1,
+            'rerun_helps': 0, 'rerun_shifts': 1, 'rerun_hurts': 0
+        },
+        {
+            'run_id': 'r3', 'status': 'FAIL', 'unknown_rate': 0.5,
+            'rerun_helps': 0, 'rerun_shifts': 0, 'rerun_hurts': 1
+        }
+    ]
+
+def test_analyze_runs_nominal_case(sample_runs_data):
+    result = core.analyze_runs(sample_runs_data, threshold=0.3, rerun_budget=1)
+    assert isinstance(result, dict)
+    assert set(result.keys()) >= {'rerun_helps', 'rerun_shifts', 'rerun_hurts'}
+    assert result['rerun_helps'] == 1
+    assert result['rerun_shifts'] == 1
+    assert result['rerun_hurts'] == 1
+
+def test_analyze_runs_empty_data():
+    result = core.analyze_runs([], threshold=0.3, rerun_budget=1)
+    assert result['rerun_helps'] == 0
+    assert result['rerun_shifts'] == 0
+    assert result['rerun_hurts'] == 0
+
+def test_analyze_runs_threshold_effect(sample_runs_data):
+    res_low = core.analyze_runs(sample_runs_data, threshold=0.1, rerun_budget=1)
+    res_high = core.analyze_runs(sample_runs_data, threshold=0.9, rerun_budget=1)
+    assert isinstance(res_low, dict)
+    assert isinstance(res_high, dict)
+    # Threshold should not break output schema
+    for key in ('rerun_helps', 'rerun_shifts', 'rerun_hurts'):
+        assert key in res_low and key in res_high
+
+def test_invalid_input_type():
+    with pytest.raises((TypeError, ValueError)):
+        core.analyze_runs('notalist', threshold=0.3, rerun_budget=1)
+
+@pytest.mark.parametrize('budget', [0, 1, 2])
+def test_valid_budget_values(sample_runs_data, budget):
+    result = core.analyze_runs(sample_runs_data, threshold=0.3, rerun_budget=budget)
+    assert isinstance(result, dict)
+    assert all(isinstance(v, int) for v in result.values())
\ No newline at end of file