@@ -5,7 +5,11 @@
 
 from plugin_eval.cli import app
 from plugin_eval.models import (
-    CompositeResult, Depth, EvalConfig, LayerResult, PluginEvalResult,
+    CompositeResult,
+    Depth,
+    EvalConfig,
+    LayerResult,
+    PluginEvalResult,
 )
 
 runner = CliRunner()
@@ -51,9 +55,7 @@ def test_plugin_eval_at_deep_depth_emits_downgrade_warning(
         assert "plugin-level" in result.stderr.lower()
         assert "deep" in result.stderr.lower()
 
-    def test_plugin_eval_at_quick_depth_does_not_warn(
-        self, sample_plugin_dir: Path
-    ) -> None:
+    def test_plugin_eval_at_quick_depth_does_not_warn(self, sample_plugin_dir: Path) -> None:
         """No warning when the requested depth is already static-only."""
         result = runner.invoke(
             app,
@@ -70,16 +72,18 @@ def test_score_warns_when_judge_unmeasured(sample_skill_dir):
         config=EvalConfig(depth=Depth.STANDARD),
         layers=[
             LayerResult(layer="static", score=0.8, sub_scores={}),
-            LayerResult(layer="judge", score=0.0, sub_scores={},
-                        metadata={"unmeasured": ["triggering_accuracy", "output_quality"]}),
+            LayerResult(
+                layer="judge",
+                score=0.0,
+                sub_scores={},
+                metadata={"unmeasured": ["triggering_accuracy", "output_quality"]},
+            ),
         ],
         composite=CompositeResult(score=60.0),
     )
     with patch("plugin_eval.cli.EvalEngine") as Eng:
         Eng.return_value.evaluate_skill.return_value = fake
-        result = CliRunner().invoke(
-            app, ["score", str(sample_skill_dir), "--output", "json"]
-        )
+        result = CliRunner().invoke(app, ["score", str(sample_skill_dir), "--output", "json"])
     assert result.exit_code == 0
     assert "judge" in result.stderr.lower()
     assert "unmeasured" in result.stderr.lower() or "could not" in result.stderr.lower()
@@ -4,8 +4,12 @@
 import pytest
 from claude_agent_sdk import AssistantMessage, ResultMessage, TextBlock
 
-from plugin_eval.layers.monte_carlo import MonteCarloAnalyzer, MonteCarloConfig, SimResult
-from plugin_eval.layers.monte_carlo import _simresult_from_messages
+from plugin_eval.layers.monte_carlo import (
+    MonteCarloAnalyzer,
+    MonteCarloConfig,
+    SimResult,
+    _simresult_from_messages,
+)
 
 
 def _assistant(text: str) -> AssistantMessage:
@@ -15,7 +19,11 @@ def _assistant(text: str) -> AssistantMessage:
 def _result(*, is_error: bool = False) -> ResultMessage:
     return ResultMessage(
         subtype="success" if not is_error else "error",
-        duration_ms=1, duration_api_ms=1, is_error=is_error, num_turns=1, session_id="t",
+        duration_ms=1,
+        duration_api_ms=1,
+        is_error=is_error,
+        num_turns=1,
+        session_id="t",
     )
 
 
@@ -66,7 +74,9 @@ def test_statistical_analysis(self):
             SimResult(activated=True, quality_score=0.8 + i * 0.002, tokens=2500, duration_ms=1200)
             for i in range(48)
         ] + [
-            SimResult(activated=False, quality_score=0.0, tokens=500, duration_ms=200, errored=True),
+            SimResult(
+                activated=False, quality_score=0.0, tokens=500, duration_ms=200, errored=True
+            ),
             SimResult(activated=True, quality_score=0.75, tokens=8000, duration_ms=5000),
         ]
         stats = analyzer._compute_statistics(results)