Implement credentialed live hardening workplan

tests/test_evaluation_scenarios.py

@@ -4,7 +4,15 @@ from pathlib import Path
 
 from phase_memory.adapters import InMemorySemanticIndex
 from phase_memory.contracts import graph_from_markitect
-from phase_memory.evaluation import EVALUATION_REPORT_SCHEMA, EVALUATION_TREND_SCHEMA, evaluation_threshold_report, evaluation_trend_artifact
+from phase_memory.evaluation import (
+    EVALUATION_REPORT_SCHEMA,
+    EVALUATION_TREND_HISTORY_SCHEMA,
+    EVALUATION_TREND_SCHEMA,
+    evaluation_threshold_report,
+    evaluation_trend_artifact,
+    load_evaluation_trend_history,
+    write_evaluation_trend_history,
+)
 from phase_memory.models import ActivationPlan, MemoryPath
 from phase_memory.retrieval import activation_quality_report, select_event_path
 from phase_memory.runtime import PhaseMemoryRuntime
@@ -126,6 +134,31 @@ def test_evaluation_trend_artifact_tracks_threshold_and_metric_deltas() -> None:
     assert trend["diagnostics"][0]["code"] == "evaluation_metric_regressed"
 
 
+def test_evaluation_trend_history_persists_without_duplicate_runs(tmp_path) -> None:
+    data = json.loads((FIXTURES / "evaluation-scenarios.json").read_text(encoding="utf-8"))
+    report = evaluation_threshold_report(data)
+    first = evaluation_trend_artifact(
+        report,
+        run_metadata={"run_id": "first", "created_at": "2026-05-19T00:00:00+00:00"},
+    )
+    second = evaluation_trend_artifact(
+        report,
+        previous_report=report,
+        run_metadata={"run_id": "second", "created_at": "2026-05-20T00:00:00+00:00"},
+    )
+    path = tmp_path / "evaluation-trend-history.json"
+
+    history = write_evaluation_trend_history(path, first)
+    history = write_evaluation_trend_history(path, first)
+    history = write_evaluation_trend_history(path, second)
+    loaded = load_evaluation_trend_history(path)
+
+    assert history["schema_version"] == EVALUATION_TREND_HISTORY_SCHEMA
+    assert loaded["count"] == 2
+    assert loaded["latest_artifact_id"] == second["id"]
+    assert "policy_denial_count" in loaded["metric_keys"]
+
+
 def _activation_plan(response):
     data = response["data"]["activation_plan"]
     return ActivationPlan(