feat(host-agent): make execution history authoritative

tests/test_host_agent_console_tasks.py

@@ -139,7 +139,11 @@ def test_authenticated_tasks_list_renders_completed_task(tmp_path) -> None:
         updated_at=datetime(2026, 7, 10, 1, tzinfo=UTC),
         completed_at=datetime(2026, 7, 10, 1, tzinfo=UTC),
     )
-    metadata_store.create_task(task)
+    metadata_store.create_task(
+        task,
+        source_task_id="cloud-task-visible",
+        source_attempt=2,
+    )
 
     client, _ = _build_client(
         tmp_path, metadata_store=metadata_store, timeline=timeline
@@ -148,7 +152,10 @@ def test_authenticated_tasks_list_renders_completed_task(tmp_path) -> None:
 
     response = client.get("/tasks")
     assert response.status_code == 200
+    assert "Executed tasks on this Host" in response.text
     assert "task-visible" in response.text
+    assert "cloud-task-visible" in response.text
+    assert "2" in response.text
     assert "completed" in response.text
 
 
@@ -190,6 +197,69 @@ def test_authenticated_task_detail_renders_timeline_with_screenshot(tmp_path) ->
     assert "find search button" in response.text
     # Screenshot inlined as base64 data URI.
     assert "data:image/png;base64," in response.text
+    assert "Before action" in response.text
+    assert "After action" in response.text
+
+
+def test_authenticated_task_detail_renders_complete_step_evidence(tmp_path) -> None:
+    from core.models import Task
+
+    metadata_store = TaskMetadataStore(tmp_path / "task_progress.sqlite3")
+    timeline = Timeline(ArtifactStore(tmp_path / "history"))
+    task = Task(
+        id="task-evidence",
+        goal="inspect search",
+        device_id="iphone-1",
+        status="completed",
+        created_at=datetime(2026, 7, 10, tzinfo=UTC),
+        updated_at=datetime(2026, 7, 10, 1, tzinfo=UTC),
+    )
+    metadata_store.create_task(
+        task,
+        source_task_id="cloud-evidence",
+        source_attempt=3,
+    )
+    timeline.append(
+        task_id=task.id,
+        scene={"screen": {"width": 10, "height": 20}, "elements": []},
+        prompt="inspect the current UI tree",
+        tool_call={"action": "get_ui_tree", "description": "inspect UI tree"},
+        result={
+            "result": [
+                {
+                    "id": "search",
+                    "type": "button",
+                    "text": "Search",
+                    "bounds": {"x": 1, "y": 2, "width": 3, "height": 4},
+                    "confidence": 0.98,
+                }
+            ]
+        },
+        before_screenshot=PNG_10X20,
+        after_screenshot=PNG_10X20 + b"after",
+        ocr_results=[
+            {
+                "text": "Search",
+                "bounds": {"x": 1, "y": 2, "width": 3, "height": 4},
+                "confidence": 0.95,
+            }
+        ],
+    )
+
+    client, _ = _build_client(
+        tmp_path, metadata_store=metadata_store, timeline=timeline
+    )
+    _login(client)
+
+    response = client.get(f"/tasks/{task.id}")
+
+    assert response.status_code == 200
+    assert "cloud-evidence" in response.text
+    assert "Cloud attempt" in response.text
+    assert "OCR results (1)" in response.text
+    assert "UI tree (1 normalized nodes)" in response.text
+    assert "inspect the current UI tree" in response.text
+    assert response.text.count("data:image/png;base64,") == 2
 
 
 def test_task_detail_404_for_unknown_task(tmp_path) -> None:
@@ -207,4 +277,4 @@ def test_tasks_list_shows_empty_state(tmp_path) -> None:
 
     response = client.get("/tasks")
     assert response.status_code == 200
-    assert "No tasks recorded" in response.text
+    assert "No executions recorded yet" in response.text