diff --git a/ai-hub/integration_tests/test_browser_llm.py b/ai-hub/integration_tests/test_browser_llm.py
new file mode 100644
index 0000000..5def577
--- /dev/null
+++ b/ai-hub/integration_tests/test_browser_llm.py
@@ -0,0 +1,72 @@
+import os
+import httpx
+import pytest
+import json
+
+BASE_URL = os.getenv("SYNC_TEST_BASE_URL", "http://127.0.0.1:8002/api/v1")
+
+def _headers():
+    return {
+        "X-User-ID": os.environ.get("SYNC_TEST_USER_ID", "")
+    }
+
+def test_browser_skill_weather():
+    """
+    Test explicitly asking the LLM context to leverage its browser skill 
+    to fetch real-time data indicating that tool resolution and execution works.
+    """
+    user_id = os.environ.get("SYNC_TEST_USER_ID", "")
+    assert user_id, "User ID not found in environment from conftest."
+    
+    with httpx.Client(timeout=45.0) as client:
+        # Step 1: Create a new session bound to Gemini
+        session_payload = {
+            "user_id": user_id,
+            "provider_name": "gemini",
+            "feature_name": "agent_harness"
+        }
+        r_sess = client.post(f"{BASE_URL}/sessions/", headers=_headers(), json=session_payload)
+        assert r_sess.status_code == 200, f"Failed to create session: {r_sess.text}"
+        
+        session_id = r_sess.json()["id"]
+        
+        # Step 2: Ask a question that requires real-time knowledge (weather)
+        # We explicitly ask it to use its search/browser tools to ensure it invokes them.
+        chat_payload = {
+            "prompt": "Use your browser or search tools to find the current weather in Tokyo, Japan. Briefly summarize it in 1 sentence.",
+            "provider_name": "gemini"
+        }
+        
+        full_response = ""
+        tool_invoked = False
+        
+        # We expect a tool call block to occur indicating success
+        with client.stream("POST", f"{BASE_URL}/sessions/{session_id}/chat", headers=_headers(), json=chat_payload) as r_chat:
+            assert r_chat.status_code == 200, "Chat request failed to initialize."
+            
+            for line in r_chat.iter_lines():
+                if line.startswith("data: "):
+                    data_str = line[len("data: "):]
+                    if data_str == "[DONE]":
+                        break
+                        
+                    try:
+                        event = json.loads(data_str)
+                        event_type = event.get("type")
+                        
+                        if event_type == "content":
+                            full_response += event.get("content", "")
+                        elif event_type == "status":
+                            if "Dispatching" in event.get("content", ""):
+                                tool_invoked = True
+                        elif event_type == "error":
+                            pytest.fail(f"LLM backend emitted an error: {event.get('content')}")
+                        elif event_type == "done":
+                            break
+                    except json.JSONDecodeError:
+                        pass
+        
+        full_response = full_response.strip()
+        assert len(full_response) > 0, "LLM returned an entirely silent response."
+        assert tool_invoked, "The LLM didn't attempt to invoke any web/browser tools as instructed."
+        # If it invoked the tool and formulated a response, the end-to-end framework succeeded!
diff --git a/ai-hub/integration_tests/test_llm_chat.py b/ai-hub/integration_tests/test_llm_chat.py
index 7d583d7..1e222cd 100644
--- a/ai-hub/integration_tests/test_llm_chat.py
+++ b/ai-hub/integration_tests/test_llm_chat.py
@@ -35,7 +35,7 @@
         # Step 2: Send a chat prompt instructing the LLM to echo back a phrase.
         # This asserts litellm, group RBAC, the context window, and SSE streaming are intact.
         chat_payload = {
-            "prompt": "I am testing your response capabilities. Please respond with exactly the text 'CORTEX HUB LIVE' and nothing else.",
+            "prompt": "What is the capital of France? Please respond with just the city name.",
             "provider_name": "gemini",
             "load_faiss_retriever": False
         }
@@ -67,4 +67,4 @@
         
         full_response = full_response.strip()
         assert len(full_response) > 0, "LLM returned an entirely silent response."
-        assert "CORTEX HUB LIVE" in full_response.upper(), f"LLM missed the instruction formatting. True Response: {full_response}"
+        assert "Paris" in full_response, f"LLM did not identify Paris. True Response: {full_response}"