diff --git a/tests/e2e/test_brain_live.py b/tests/e2e/test_brain_live.py
new file mode 100644
index 0000000..a9de297
--- /dev/null
+++ b/tests/e2e/test_brain_live.py
@@ -0,0 +1,34 @@
+import pytest
+from GramAddict.core.navigation.brain import ask_brain_for_action
+from GramAddict.core.perception.screen_identity import ScreenType
+import logging
+
+logger = logging.getLogger(__name__)
+
+@pytest.mark.live_llm
+def test_brain_recommends_scroll_when_trapped():
+    """
+    Test that the real, live LLM Brain correctly deduces that it should 
+    scroll down when the target element is missing and it's trapped.
+    """
+    goal = "open following list"
+    screen = "OWN_PROFILE"
+    available_actions = ["tap profile tab", "tap share button", "press back", "tap reels tab", "tap messages tab", "scroll down", "scroll up"]
+    explored_nav_actions = {"tap following list"}
+    
+    # We query the actual LLM as configured in the environment (e.g. qwen3.5:latest)
+    # This prevents regressions where the LLM is misconfigured or returns empty strings.
+    brain_action = ask_brain_for_action(
+        goal=goal,
+        screen_type=screen,
+        available_actions=available_actions,
+        explored_actions=explored_nav_actions
+    )
+    
+    logger.info(f"Brain action returned: '{brain_action}'")
+    
+    assert brain_action is not None, "Brain LLM returned None. Is the URL/Model configured correctly?"
+    assert brain_action != "", "Brain LLM returned an empty string."
+    
+    # The brain should reasonably choose 'scroll down' to find the missing following list
+    assert brain_action == "scroll down", f"Expected Brain to choose 'scroll down', but got '{brain_action}'"