test: add pure-active session output-token double-count coverage (#1155)

tests/copilot_usage/test_render_detail.py

@@ -850,6 +850,37 @@ def test_active_session_shows_active_period_panel(self) -> None:
         output = _strip_ansi(buf.getvalue())
         assert "Active Period" in output
 
+    def test_pure_active_session_aggregate_stats_token_count(self) -> None:
+        """render_session_detail Aggregate Stats panel shows correct (non-doubled) tokens."""
+        summary = SessionSummary(
+            session_id="pure-active-e2e-tok",
+            start_time=datetime(2026, 4, 1, 10, 0, 0, tzinfo=UTC),
+            is_active=True,
+            has_shutdown_metrics=False,
+            model_calls=2,
+            user_messages=1,
+            active_model_calls=2,
+            active_user_messages=1,
+            active_output_tokens=500,
+            model_metrics={
+                "claude-sonnet-4": ModelMetrics(
+                    usage=TokenUsage(outputTokens=500),
+                )
+            },
+        )
+        ev = SessionEvent(
+            type=EventType.USER_MESSAGE,
+            timestamp=datetime(2026, 4, 1, 10, 5, 0, tzinfo=UTC),
+            data={"content": "test"},
+        )
+        buf, console = _buf_console()
+        render_session_detail([ev], summary, target_console=console)
+        output = _strip_ansi(buf.getvalue())
+
+        assert "Aggregate Stats" in output
+        assert "500" in output
+        assert "1000" not in output  # double-count must not appear
+
 
 # ---------------------------------------------------------------------------
 # _event_type_label — parametrized unit tests (issue #879)

tests/copilot_usage/test_report.py

@@ -939,6 +939,33 @@ def test_api_duration_formatted(self) -> None:
         output = _capture_console(_render_aggregate_stats, summary)
         assert "1m" in output
 
+    def test_pure_active_session_output_tokens_not_double_counted(self) -> None:
+        """Pure-active session: Aggregate Stats shows model_metrics tokens, not 2× that value."""
+        summary = SessionSummary(
+            session_id="pure-active-agg",
+            is_active=True,
+            has_shutdown_metrics=False,
+            model_calls=4,
+            user_messages=3,
+            active_model_calls=4,
+            active_user_messages=3,
+            active_output_tokens=800,
+            model_metrics={
+                "gpt-4": ModelMetrics(
+                    usage=TokenUsage(outputTokens=800),
+                )
+            },
+        )
+
+        output = _capture_console(_render_aggregate_stats, summary)
+
+        # Should show 800, NOT 1600 (double-count)
+        assert "800" in output, "Expected 800 output tokens in Aggregate Stats"
+        assert "1600" not in output, "Double-counted token value must not appear"
+        assert (
+            "1.6K" not in output
+        ), "Double-counted token value must not appear as formatted K value"
+
 
 # ---------------------------------------------------------------------------
 # format_tokens tests