feat(examples): add Pydantic AI zero-code OTLP example

examples/README.md

@@ -29,6 +29,7 @@ agentevals accepts OTLP/HTTP on port 4318 (`http/protobuf` and `http/json`) and
 | [zero-code-examples/ollama/](./zero-code-examples/ollama/) | LangChain | Ollama |
 | [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 |
 
 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).
 
@@ -103,6 +104,7 @@ Detection checks for `gen_ai.request.model` / `gen_ai.input.messages` (GenAI sem
 | [zero-code-examples/ollama/](./zero-code-examples/ollama/) | LangChain | Ollama | GenAI semconv (logs) | Standard OTLP export |
 | [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 |
 | [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 |
@@ -217,6 +219,7 @@ python examples/zero-code-examples/langchain/run.py
 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
 
 # SDK examples:
 python examples/sdk_example/context_manager_example.py
@@ -232,7 +235,7 @@ python examples/strands_agent/main.py
 Traces stream to the dev server in real-time. Evaluation runs automatically when the session completes.
 
 See each example's README for prerequisites and detailed instructions:
-- [zero-code-examples/](./zero-code-examples/) (LangChain + Strands, standard OTLP)
+- [zero-code-examples/](./zero-code-examples/) (LangChain, Strands, ADK, OpenAI Agents, Pydantic AI — standard OTLP)
 - [dice_agent/README.md](./dice_agent/README.md) (Google ADK + Gemini)
 - [langchain_agent/README.md](./langchain_agent/README.md) (LangChain + OpenAI, SDK)
 - [strands_agent/](./strands_agent/) (Strands + OpenAI, SDK)

examples/zero-code-examples/pydantic-ai/requirements.txt

@@ -0,0 +1,5 @@
+pydantic-ai>=1.81.0
+
+opentelemetry-sdk>=1.36.0
+opentelemetry-exporter-otlp-proto-http>=1.36.0
+python-dotenv>=1.0.0

examples/zero-code-examples/pydantic-ai/run.py

@@ -0,0 +1,105 @@
+"""Run a dice-rolling Pydantic AI agent with OTLP export — no agentevals SDK.
+
+Demonstrates zero-code integration: any OTel-instrumented agent streams
+traces to agentevals by pointing the OTLP exporter at the receiver.
+
+Pydantic AI has built-in OTel support via Agent.instrument_all(). By default
+it uses version 2 of the GenAI semconv format, storing message content in span
+attributes — only a TracerProvider is needed.
+No separate instrumentation library is needed.
+
+Prerequisites:
+    1. pip install -r requirements.txt
+    2. agentevals serve --dev
+    3. export OPENAI_API_KEY="your-key-here"
+
+Usage:
+    python examples/zero-code-examples/pydantic-ai/run.py
+"""
+
+import os
+import random
+
+from dotenv import load_dotenv
+from opentelemetry import trace
+from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
+from opentelemetry.sdk.resources import Resource
+from opentelemetry.sdk.trace import TracerProvider
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from pydantic_ai import Agent
+
+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
+    for i in range(2, int(number**0.5) + 1):
+        if number % i == 0:
+            return False
+    return True
+
+
+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=pydantic_ai_eval,agentevals.session_name=pydantic-ai-zero-code",
+    )
+
+    resource = Resource.create()
+
+    tracer_provider = TracerProvider(resource=resource)
+    tracer_provider.add_span_processor(BatchSpanProcessor(OTLPSpanExporter(), schedule_delay_millis=1000))
+    trace.set_tracer_provider(tracer_provider)
+
+    # Enable Pydantic AI's built-in OTel instrumentation. This one call
+    # wires up all agents globally — no framework-specific instrumentor
+    # library (like opentelemetry-instrumentation-openai-v2) is needed.
+    Agent.instrument_all()
+
+    agent = Agent(
+        "openai:gpt-4o-mini",
+        instructions="You are a helpful assistant. You can roll dice and check if numbers are prime.",
+    )
+    agent.tool_plain(roll_die)
+    agent.tool_plain(check_prime)
+
+    test_queries = [
+        "Hi! Can you help me?",
+        "Roll a 20-sided die for me",
+        "Is the number you rolled prime?",
+    ]
+
+    message_history = []
+
+    try:
+        for i, query in enumerate(test_queries, 1):
+            print(f"\n[{i}/{len(test_queries)}] User: {query}")
+
+            result = agent.run_sync(query, message_history=message_history)
+
+            print(f"     Agent: {result.output}")
+
+            # Pass the full message history forward for multi-turn conversation.
+            message_history = result.all_messages()
+    finally:
+        print()
+        tracer_provider.force_flush()
+        print("All traces flushed to OTLP receiver.")
+
+
+if __name__ == "__main__":
+    main()

tests/integration/test_live_agents.py

@@ -305,6 +305,66 @@ def test_session_visible_via_api(self, live_servers):
         assert session_name in session_ids
 
 
+@_skip_no_openai
+class TestPydanticAIZeroCode:
+    """Run the Pydantic AI zero-code OTLP example and verify session grouping."""
+
+    def test_session_created_with_spans(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-pydantic-ai"
+
+        result = _run_agent(
+            "examples/zero-code-examples/pydantic-ai/run.py",
+            otlp_http_port,
+            session_name,
+        )
+        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 LLM calls"
+
+    def test_invocations_extracted_with_content(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-pydantic-ai-inv"
+
+        result = _run_agent(
+            "examples/zero-code-examples/pydantic-ai/run.py",
+            otlp_http_port,
+            session_name,
+        )
+        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"
+        for inv in session.invocations:
+            has_content = inv.get("userText") or inv.get("agentResponse")
+            assert has_content, f"Invocation {inv.get('invocationId', '?')} has no content"
+
+    def test_session_visible_via_api(self, live_servers):
+        main_port, otlp_http_port, mgr = live_servers
+        session_name = "e2e-pydantic-ai-api"
+
+        result = _run_agent(
+            "examples/zero-code-examples/pydantic-ai/run.py",
+            otlp_http_port,
+            session_name,
+        )
+        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