diff --git a/literalai/instrumentation/llamaindex.py b/literalai/instrumentation/llamaindex.py
index 09ce4b4..0b9fbb0 100644
--- a/literalai/instrumentation/llamaindex.py
+++ b/literalai/instrumentation/llamaindex.py
@@ -1,7 +1,7 @@
 import logging
 import uuid
-from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypedDict, Union, cast
-
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, Union, cast
+from typing_extensions import TypedDict
 from llama_index.core.base.llms.types import MessageRole
 from llama_index.core.base.response.schema import Response, StreamingResponse
 from llama_index.core.instrumentation import get_dispatcher
@@ -30,7 +30,7 @@
 from llama_index.core.schema import NodeWithScore, QueryBundle, TextNode
 from openai.types import CompletionUsage
 from openai.types.chat import ChatCompletion
-from pydantic import Field
+from pydantic import PrivateAttr
 
 from literalai.context import active_thread_var
 from literalai.observability.generation import ChatGeneration, GenerationMessageRole
@@ -111,8 +111,8 @@ def create_generation(event: LLMChatStartEvent):
 class LiteralEventHandler(BaseEventHandler):
     """This class handles events coming from LlamaIndex."""
 
-    _client: "LiteralClient" = Field(...)
-    _span_handler: "LiteralSpanHandler" = Field(...)
+    _client: "LiteralClient" = PrivateAttr(...)
+    _span_handler: "LiteralSpanHandler" = PrivateAttr(...)
     runs: Dict[str, List[Step]] = {}
     streaming_run_ids: List[str] = []
 
diff --git a/tests/e2e/test_llamaindex.py b/tests/e2e/test_llamaindex.py
new file mode 100644
index 0000000..bee2652
--- /dev/null
+++ b/tests/e2e/test_llamaindex.py
@@ -0,0 +1,42 @@
+import os
+import urllib.parse
+
+import pytest
+
+from literalai import LiteralClient
+
+from dotenv import load_dotenv
+
+load_dotenv()
+
+
+@pytest.fixture
+def non_mocked_hosts() -> list:
+    non_mocked_hosts = []
+
+    # Always skip mocking API
+    url = os.getenv("LITERAL_API_URL", None)
+    if url is not None:
+        parsed = urllib.parse.urlparse(url)
+        non_mocked_hosts.append(parsed.hostname)
+
+    return non_mocked_hosts
+
+
+@pytest.mark.e2e
+class TestLlamaIndex:
+    @pytest.fixture(
+        scope="class"
+    )  # Feel free to move this fixture up for further testing
+    def client(self):
+        url = os.getenv("LITERAL_API_URL", None)
+        api_key = os.getenv("LITERAL_API_KEY", None)
+        assert url is not None and api_key is not None, "Missing environment variables"
+
+        client = LiteralClient(batch_size=5, url=url, api_key=api_key)
+        client.instrument_llamaindex()
+
+        return client
+
+    async def test_instrument_llamaindex(self, client: "LiteralClient"):
+        assert client is not None