diff --git a/backend/app/code/eval/scoring.py b/backend/app/code/eval/scoring.py
index 508f7da1..7e414198 100644
--- a/backend/app/code/eval/scoring.py
+++ b/backend/app/code/eval/scoring.py
@@ -51,8 +51,8 @@ def _score_static(self, result: StaticEvalResult) -> Dict[str, float]:
         scores["syntax"] = 100.0 if result.syntax_valid else 0.0
         
         # Risk score: Deduct for each risk found
-        risk_penalty = min(len(result.risks) * 10, 50)
-        scores["risks"] = max(100.0 - risk_penalty, 50.0)
+        risk_penalty = min(len(result.risks) * 10, 100)
+        scores["risks"] = max(100.0 - risk_penalty, 0.0)
         
         # Lint score: Deduct for warnings
         warning_penalty = min(result.warning_count * 5, 30)
diff --git a/backend/tests/test_pr_13_scoring_risk_floor.py b/backend/tests/test_pr_13_scoring_risk_floor.py
new file mode 100644
index 00000000..e0681bb7
--- /dev/null
+++ b/backend/tests/test_pr_13_scoring_risk_floor.py
@@ -0,0 +1,65 @@
+"""
+PR-B13: risk score no longer floored at 50.
+
+With 10+ risks the score should drop to 0, not stay at 50.
+With 0 risks the score should be 100.
+"""
+
+import sys
+import os
+import pytest
+
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+
+from app.code.eval.scoring import EvalScorer
+from app.code.eval.static_eval import StaticEvalResult
+
+
+def _make_result(risks_count):
+    return StaticEvalResult(
+        passed=True,
+        syntax_valid=True,
+        error_count=0,
+        warning_count=0,
+        info_count=0,
+        diagnostics=[],
+        risks=[{"type": "test_risk"} for _ in range(risks_count)],
+    )
+
+
+class TestScoringRiskFloor:
+
+    def setup_method(self):
+        self.scorer = EvalScorer()
+
+    def test_no_risks_gives_full_score(self):
+        result = _make_result(0)
+        scores = self.scorer._score_static(result)
+        assert scores["risks"] == 100.0
+
+    def test_three_risks_gives_70(self):
+        result = _make_result(3)
+        scores = self.scorer._score_static(result)
+        assert scores["risks"] == 70.0
+
+    def test_five_risks_gives_50(self):
+        result = _make_result(5)
+        scores = self.scorer._score_static(result)
+        assert scores["risks"] == 50.0
+
+    def test_ten_risks_gives_zero(self):
+        result = _make_result(10)
+        scores = self.scorer._score_static(result)
+        assert scores["risks"] == 0.0
+
+    def test_many_risks_clamped_at_zero(self):
+        result = _make_result(50)
+        scores = self.scorer._score_static(result)
+        assert scores["risks"] == 0.0
+
+    def test_one_risk_distinguishable_from_many(self):
+        """The old floor made 1 risk and 50 risks look the same (both 50)."""
+        score_one = self.scorer._score_static(_make_result(1))["risks"]
+        score_many = self.scorer._score_static(_make_result(50))["risks"]
+        assert score_one != score_many
+        assert score_one > score_many