diff --git a/frontend/components/best-practices/AnalysisHydration.tsx b/frontend/components/best-practices/AnalysisHydration.tsx
index 004df2ee..a09786e0 100644
--- a/frontend/components/best-practices/AnalysisHydration.tsx
+++ b/frontend/components/best-practices/AnalysisHydration.tsx
@@ -24,7 +24,7 @@ export function AnalysisHydration({
     partial.best_practices_score != null;
 
   return (
-    <div className="flex h-full flex-col space-y-4">
+    <div className="flex flex-col space-y-4">
       <div className="flex items-center gap-2 text-sm font-semibold text-foreground">
         <Loader2 size={15} className="animate-spin text-purple-300" />
         Generating analysis
diff --git a/frontend/components/best-practices/AnalysisModal.tsx b/frontend/components/best-practices/AnalysisModal.tsx
index bafa945b..6a678add 100644
--- a/frontend/components/best-practices/AnalysisModal.tsx
+++ b/frontend/components/best-practices/AnalysisModal.tsx
@@ -342,7 +342,7 @@ export function AnalysisModal({
             </div>
 
             {/* RIGHT: Telemetry & Tabbed Analysis */}
-            <div className="flex min-h-0 flex-col overflow-hidden bg-[#141414] p-5">
+            <div className="flex min-h-0 flex-col overflow-y-auto bg-[#141414] p-5">
               <ExplainPanel
                 explanation={explanation}
                 loading={loading}
diff --git a/frontend/components/best-practices/ExplainPanel.tsx b/frontend/components/best-practices/ExplainPanel.tsx
index d2728bc4..13798897 100644
--- a/frontend/components/best-practices/ExplainPanel.tsx
+++ b/frontend/components/best-practices/ExplainPanel.tsx
@@ -48,7 +48,7 @@ export function ExplainPanel({
   if (!explanation) return null;
 
   return (
-    <div className="flex h-full flex-col space-y-4">
+    <div className="flex flex-col space-y-4">
       {/* Top Telemetry: Quality Gauge + Score Radar */}
       <div className="grid shrink-0 grid-cols-2 gap-3">
         <QualityGauge score={explanation.quality_score} />
@@ -80,7 +80,7 @@ export function ExplainPanel({
       </div>
 
       {/* Scrollable Tab Content */}
-      <div className="min-h-0 flex-1 space-y-4 overflow-y-auto pr-1">
+      <div className="space-y-4 pr-1">
         {activeTab === "overview" && (
           <>
             <AnalysisSummary summary={explanation.summary} />
diff --git a/internal/executor/executor.go b/internal/executor/executor.go
index b18e3e65..edf8e299 100644
--- a/internal/executor/executor.go
+++ b/internal/executor/executor.go
@@ -983,6 +983,23 @@ func formatPythonLiteral(paramType string, data []byte) (string, error) {
 	return "", fmt.Errorf("unsupported type %q for Python literal", paramType)
 }
 
+// formatPythonExpected ensures a test case expected value is valid JSON
+// before it is embedded in the Python test template. Bare strings (e.g.
+// fish) are wrapped as JSON strings ("fish") so json.loads in the template
+// does not fail with Expecting value.
+func formatPythonExpected(raw string) string {
+	var js json.RawMessage
+	if err := json.Unmarshal([]byte(raw), &js); err == nil {
+		return raw // already valid JSON
+	}
+	// Not valid JSON — treat as a bare string and wrap it as a JSON string.
+	b, err := json.Marshal(raw)
+	if err != nil {
+		return raw // should never happen for a Go string
+	}
+	return string(b)
+}
+
 // executePython runs Python code execution: prepares sandbox files,
 // executes, parses output, optionally records submission + progress, and returns the result.
 func (e *Executor) executePython(ctx context.Context, req ExecutionRequest, problem *store.Problem, testCases []store.TestCase, recordSubmission bool) (*ExecutionResult, error) {
@@ -1019,7 +1036,7 @@ func (e *Executor) executePython(ctx context.Context, req ExecutionRequest, prob
 		pyCases[i] = PyTestCaseRenderData{
 			Ordinal:  tc.Ordinal,
 			PyInputs: pyInputs,
-			Expected: tc.Expected,
+			Expected: formatPythonExpected(tc.Expected),
 		}
 	}
 
diff --git a/internal/executor/executor_test.go b/internal/executor/executor_test.go
index 5e261812..844ed9e9 100644
--- a/internal/executor/executor_test.go
+++ b/internal/executor/executor_test.go
@@ -369,6 +369,37 @@ func TestPythonTemplate_ComparisonLogic(t *testing.T) {
 	if !strings.Contains(content, "result == expected_val") {
 		t.Error("template missing result == expected_val")
 	}
+	if !strings.Contains(content, "json.JSONDecodeError") {
+		t.Error("template missing JSONDecodeError fallback for non-JSON expected values")
+	}
+}
+
+func TestFormatPythonExpected(t *testing.T) {
+	tests := []struct {
+		raw  string
+		want string
+	}{
+		// Valid JSON passthrough
+		{"42", "42"},
+		{"true", "true"},
+		{"null", "null"},
+		{"[1,2,3]", "[1,2,3]"},
+		{`"hello"`, `"hello"`},
+		// Bare strings wrapped as JSON strings
+		{"fish", `"fish"`},
+		{"hello world", `"hello world"`},
+		{"", `""`},
+		// Strings with special chars
+		{"it's", `"it's"`},
+		{`he said "hi"`, `"he said \"hi\""`},
+	}
+
+	for _, tc := range tests {
+		got := formatPythonExpected(tc.raw)
+		if got != tc.want {
+			t.Errorf("formatPythonExpected(%q) = %q, want %q", tc.raw, got, tc.want)
+		}
+	}
 }
 
 func TestResolveProblemLanguageMeta(t *testing.T) {
diff --git a/internal/executor/templates.go b/internal/executor/templates.go
index 68439e1c..5550926c 100644
--- a/internal/executor/templates.go
+++ b/internal/executor/templates.go
@@ -53,7 +53,10 @@ for tc in test_cases:
     expected = tc["expected"]
     try:
         result = {{.FuncName}}(*inputs)
-        expected_val = json.loads(expected)
+        try:
+            expected_val = json.loads(expected)
+        except json.JSONDecodeError:
+            expected_val = expected
         if result == expected_val:
             passed += 1
             print(f"--- PASS: TestSolution/case_{ordinal}")