Harden golden QA readiness gate

backend/README.md

@@ -310,6 +310,12 @@ Machine-readable output:
 python scripts/run_golden_qa_benchmark.py --json
 ```
 
+Shell wrapper used by release-readiness checks:
+
+```bash
+bash scripts/verify_golden_qa_benchmark.sh
+```
+
 The benchmark compares `fixtures/golden/predicted_buildings.geojson` against `fixtures/golden/reference_buildings.geojson` and fails on metric drift. Expected baseline:
 
 - precision: `0.5`
@@ -321,6 +327,10 @@ The benchmark compares `fixtures/golden/predicted_buildings.geojson` against `fi
 
 The command uses existing QA/QC service logic and verifies `QualityCheck`/`Metric` persistence through an in-memory test session. It does not require live providers, AI models, Docker or PostGIS.
 
+`scripts/run_readiness_check.sh` runs this benchmark automatically, so any
+change that alters the golden QA/QC metric baseline must update the fixture and
+expected metrics deliberately.
+
 ### Demo workflow seed
 
 Sprint 15 adds an explicit offline demo workflow seed. It creates or returns a

backend/tests/test_readiness_gate.py

@@ -22,6 +22,14 @@ def test_readiness_gate_checks_demo_export_workflow_script_syntax() -> None:
     assert "bash -n scripts/verify_demo_export_workflow.sh" in content
 
 
+def test_readiness_gate_runs_golden_qa_benchmark() -> None:
+    script = Path(__file__).resolve().parents[2] / "scripts" / "run_readiness_check.sh"
+    content = script.read_text(encoding="utf-8")
+
+    assert "scripts/run_golden_qa_benchmark.py --json" in content
+    assert "bash -n scripts/verify_golden_qa_benchmark.sh" in content
+
+
 def test_readiness_gate_compiles_demo_cleanup_script() -> None:
     script = Path(__file__).resolve().parents[2] / "scripts" / "run_readiness_check.sh"
     content = script.read_text(encoding="utf-8")

backend/tests/test_sprint12_golden_qa_benchmark.py

@@ -55,3 +55,20 @@ def test_golden_qa_benchmark_command_passes_and_reports_persistence() -> None:
         "precision",
         "recall",
     ]
+
+
+def test_golden_qa_shell_wrapper_is_safe_and_documented() -> None:
+    script = ROOT / "scripts" / "verify_golden_qa_benchmark.sh"
+    content = script.read_text(encoding="utf-8")
+
+    assert "set -euo pipefail" in content
+    assert "run_golden_qa_benchmark.py --json" in content
+
+    result = subprocess.run(
+        ["bash", "-n", "scripts/verify_golden_qa_benchmark.sh"],
+        cwd=ROOT,
+        check=True,
+        text=True,
+        capture_output=True,
+    )
+    assert result.returncode == 0