feat(examples): add LlamaIndex zero-code OTLP example

Uses official LlamaIndexOpenTelemetry integration. Adds e2e tests.

Author: mesutoezdil

examples/README.md

@@ -30,6 +30,7 @@ agentevals accepts OTLP/HTTP on port 4318 (`http/protobuf` and `http/json`) and
 | [zero-code-examples/strands/](./zero-code-examples/strands/) | Strands | OpenAI |
 | [zero-code-examples/adk/](./zero-code-examples/adk/) | Google ADK | Gemini |
 | [zero-code-examples/pydantic-ai/](./zero-code-examples/pydantic-ai/) | Pydantic AI | OpenAI |
+| [zero-code-examples/llama-index/](./zero-code-examples/llama-index/) | LlamaIndex | OpenAI |
 
 This approach works with any framework that has OTel instrumentation: LangChain, Strands, Google ADK, etc. If your framework already emits OTel spans, you only need to add `OTLPSpanExporter` (and `OTLPLogExporter` if it uses GenAI log-based content delivery).
 
@@ -105,6 +106,7 @@ Detection checks for `gen_ai.request.model` / `gen_ai.input.messages` (GenAI sem
 | [zero-code-examples/strands/](./zero-code-examples/strands/) | Strands | OpenAI | GenAI semconv (events*) | Standard OTLP export |
 | [zero-code-examples/adk/](./zero-code-examples/adk/) | Google ADK | Gemini | ADK built-in | Standard OTLP export |
 | [zero-code-examples/pydantic-ai/](./zero-code-examples/pydantic-ai/) | Pydantic AI | OpenAI | GenAI semconv (span attrs) | Standard OTLP export |
+| [zero-code-examples/llama-index/](./zero-code-examples/llama-index/) | LlamaIndex | OpenAI | GenAI semconv (logs) | Standard OTLP export |
 | [langchain_agent](./langchain_agent/) | LangChain | OpenAI | GenAI semconv (logs) | SDK WebSocket |
 | [strands_agent](./strands_agent/) | Strands | OpenAI | GenAI semconv (events*) | SDK WebSocket |
 | [dice_agent](./dice_agent/) | Google ADK | Gemini | ADK built-in | SDK WebSocket |
@@ -226,6 +228,7 @@ python examples/zero-code-examples/ollama/run.py
 python examples/zero-code-examples/strands/run.py
 python examples/zero-code-examples/adk/run.py
 python examples/zero-code-examples/pydantic-ai/run.py
+python examples/zero-code-examples/llama-index/run.py
 
 # SDK examples:
 python examples/sdk_example/context_manager_example.py

examples/zero-code-examples/llama-index/requirements.txt

@@ -0,0 +1,7 @@
+llama-index-core>=0.14.0
+llama-index-llms-openai-like>=0.3.0
+llama-index-observability-otel>=0.1.0
+
+opentelemetry-sdk>=1.36.0
+opentelemetry-exporter-otlp-proto-http>=1.36.0
+python-dotenv>=1.0.0

examples/zero-code-examples/llama-index/run.py

@@ -0,0 +1,98 @@
+"""Run a LlamaIndex dice agent with standard OTLP export.
+
+Uses the official LlamaIndexOpenTelemetry integration to set up OTel tracing.
+It handles the tracer provider setup and span export internally.
+Traces stream to agentevals via OTLPSpanExporter with no agentevals SDK needed.
+
+Note: LlamaIndexOpenTelemetry exports spans only. Log-based content delivery
+is not part of this integration. Message content is captured via span attributes.
+
+Prerequisites:
+    1. pip install -r examples/zero-code-examples/llama-index/requirements.txt
+    2. agentevals serve --dev
+    3. export OPENAI_API_KEY="your-key-here"
+
+Usage:
+    python examples/zero-code-examples/llama-index/run.py
+"""
+
+import asyncio
+import os
+import random
+
+from dotenv import load_dotenv
+from llama_index.core.agent.workflow import FunctionAgent
+from llama_index.core.tools import FunctionTool
+from llama_index.llms.openai_like import OpenAILike
+from llama_index.observability.otel import LlamaIndexOpenTelemetry
+from opentelemetry import trace
+from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
+from opentelemetry.sdk.resources import Resource
+
+load_dotenv(override=True)
+
+
+def roll_die(sides: int) -> int:
+    """Roll a die with the given number of sides and return the result."""
+    return random.randint(1, sides)
+
+
+def check_prime(number: int) -> bool:
+    """Return True if the number is prime, False otherwise."""
+    if number < 2:
+        return False
+    return all(number % i for i in range(2, int(number**0.5) + 1))
+
+
+async def main():
+    if not os.getenv("OPENAI_API_KEY"):
+        print("OPENAI_API_KEY not set.")
+        return
+
+    endpoint = os.environ.get("OTEL_EXPORTER_OTLP_ENDPOINT", "http://localhost:4318")
+    print(f"OTLP endpoint: {endpoint}")
+
+    os.environ.setdefault(
+        "OTEL_RESOURCE_ATTRIBUTES",
+        "agentevals.eval_set_id=llama_index_eval,agentevals.session_name=llama-index-zero-code",
+    )
+
+    resource = Resource.create()
+
+    LlamaIndexOpenTelemetry(
+        span_exporter=OTLPSpanExporter(),
+        span_processor="batch",
+        service_name_or_resource=resource,
+    ).start_registering()
+
+    llm = OpenAILike(
+        model=os.environ.get("OPENAI_MODEL", "gpt-4o-mini"),
+        api_base=os.environ.get("OPENAI_BASE_URL", "https://api.openai.com/v1"),
+        is_chat_model=True,
+        is_function_calling_model=True,
+    )
+    agent = FunctionAgent(
+        tools=[FunctionTool.from_defaults(fn=roll_die), FunctionTool.from_defaults(fn=check_prime)],
+        llm=llm,
+        system_prompt="You are a helpful assistant. You can roll dice and check if numbers are prime.",
+    )
+
+    test_queries = [
+        "Hi! Can you help me?",
+        "Roll a 20-sided die for me",
+        "Is the number you rolled prime?",
+    ]
+
+    try:
+        for i, query in enumerate(test_queries, 1):
+            print(f"\n[{i}/{len(test_queries)}] User: {query}")
+            result = await agent.run(query)
+            print(f"     Agent: {result.response.content}")
+    finally:
+        print()
+        trace.get_tracer_provider().force_flush()
+        print("All traces flushed to OTLP receiver.")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

tests/integration/test_live_agents.py

@@ -365,6 +365,72 @@ def test_session_visible_via_api(self, live_servers):
         assert session_name in session_ids
 
 
+@_skip_no_openai
+class TestLlamaIndexZeroCode:
+    """Run the LlamaIndex zero-code OTLP example and verify session grouping."""
+
+    def test_session_created_spans_only(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-llama-index"
+
+        result = _run_agent(
+            "examples/zero-code-examples/llama-index/run.py",
+            otlp_http_port,
+            session_name,
+            extra_env={
+                "OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT": "true",
+            },
+        )
+        assert result.returncode == 0, f"Agent failed:\nstdout: {result.stdout}\nstderr: {result.stderr}"
+
+        wait_for_session_complete_sync(mgr, session_name, timeout=30)
+        session = mgr.sessions[session_name]
+
+        assert session.is_complete
+        assert session.source == "otlp"
+        assert len(session.spans) > 0, "Expected spans from LlamaIndex agent"
+
+    def test_invocations_extracted(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-llama-index-inv"
+
+        result = _run_agent(
+            "examples/zero-code-examples/llama-index/run.py",
+            otlp_http_port,
+            session_name,
+            extra_env={
+                "OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT": "true",
+            },
+        )
+        assert result.returncode == 0, f"Agent failed:\nstdout: {result.stdout}\nstderr: {result.stderr}"
+
+        wait_for_session_complete_sync(mgr, session_name, timeout=30)
+        session = mgr.sessions[session_name]
+
+        assert len(session.invocations) > 0, "Expected extracted invocations"
+
+    def test_session_visible_via_api(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-llama-index-api"
+
+        result = _run_agent(
+            "examples/zero-code-examples/llama-index/run.py",
+            otlp_http_port,
+            session_name,
+            extra_env={
+                "OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT": "true",
+            },
+        )
+        assert result.returncode == 0
+
+        wait_for_session_complete_sync(mgr, session_name, timeout=30)
+
+        resp = httpx.get(f"http://127.0.0.1:{main_port}/api/streaming/sessions")
+        assert resp.status_code == 200
+        session_ids = [s["sessionId"] for s in resp.json()["data"]]
+        assert session_name in session_ids
+
+
 @_skip_no_openai
 class TestAgentRerun:
     """Verify that re-running an agent with the same session_name creates