loop-benchmarking

commit 4c1ca47d2ad560fc2037407b1b702e3ec33d9497
parent afa692e0d3097240748b70e991a55fe00a7fb034
Author: Brian Graham <brian@buildingbetterteams.de>
Date:   Mon,  6 Apr 2026 10:44:17 +0200

Add sonarqube metric to analysis pipeline, fix metric labels

- experiment_design.py: added sonarqube and build_quality extractors
- clean-and-reeval.py: added sonarqube and build_quality to analysis metrics
- Insights.tsx: metric labels updated (Outcome, Code Quality = SonarQube,
  Build Quality = lint/typecheck, Code Analysis = our homebrew)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Diffstat:
M
dashboard/src/components/Insights.tsx
 | 
7
++++---
M
harness/clean-and-reeval.py
 | 
3
++-
M
harness/lib/experiment_design.py
 | 
12
++++++++++++

3 files changed, 18 insertions(+), 4 deletions(-)
diff --git a/dashboard/src/components/Insights.tsx b/dashboard/src/components/Insights.tsx
@@ -9,11 +9,12 @@ interface InsightsProps {
 }
 
 const METRICS = [
-  { key: "score", label: "Overall" },
+  { key: "score", label: "Outcome" },
   { key: "gameplay", label: "Gameplay" },
-  { key: "code_quality", label: "Code" },
+  { key: "sonarqube", label: "Code Quality" },
+  { key: "quality", label: "Build Quality" },
   { key: "structural", label: "Structural" },
-  { key: "quality", label: "Lint/Types" },
+  { key: "code_quality", label: "Code Analysis" },
   { key: "transcript", label: "Agent Eff." },
   { key: "cost", label: "Cost" },
   { key: "turns", label: "Turns" },
diff --git a/harness/clean-and-reeval.py b/harness/clean-and-reeval.py
@@ -117,7 +117,8 @@ def run_analysis():
 
     metrics = [
         "score", "cost", "turns", "wall_time",
-        "gameplay", "code_quality", "structural", "transcript",
+        "gameplay", "sonarqube", "code_quality",
+        "structural", "transcript", "build_quality",
     ]
     for metric in metrics:
         effects = analyze_main_effects(str(RESULTS_DIR), metric)
diff --git a/harness/lib/experiment_design.py b/harness/lib/experiment_design.py
@@ -535,6 +535,18 @@ def _extract_metric(run, metric):
             val = t.get("score")
             return val if isinstance(val, (int, float)) else None
         return None
+    elif metric == "sonarqube":
+        sq = run["eval"].get("sonarqube", {})
+        if isinstance(sq, dict):
+            val = sq.get("score")
+            return val if isinstance(val, (int, float)) else None
+        return None
+    elif metric == "build_quality":
+        q = run["eval"].get("quality", {})
+        if isinstance(q, dict):
+            val = q.get("score")
+            return val if isinstance(val, (int, float)) else None
+        return None
     return None