feat(examples): add AWS AgentCore zero-code OTLP example

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/agentcore/](./zero-code-examples/agentcore/) | AWS AgentCore | Amazon Bedrock |
 
 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/agentcore/](./zero-code-examples/agentcore/) | AWS AgentCore | Amazon Bedrock | GenAI semconv (events*) | 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 |
@@ -220,6 +222,7 @@ 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/agentcore/run.py
 
 # SDK examples:
 python examples/sdk_example/context_manager_example.py
@@ -235,7 +238,7 @@ 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, ADK, OpenAI Agents, Pydantic AI — standard OTLP)
+- [zero-code-examples/](./zero-code-examples/) (LangChain, Strands, ADK, OpenAI Agents, Pydantic AI, AWS AgentCore, 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/agentcore/requirements.txt

@@ -0,0 +1,5 @@
+strands-agents>=1.35.0
+boto3>=1.38.0
+opentelemetry-sdk>=1.36.0
+opentelemetry-exporter-otlp-proto-http>=1.36.0
+python-dotenv>=1.0.0

examples/zero-code-examples/agentcore/run.py

@@ -0,0 +1,54 @@
+"""AWS AgentCore zero-code OTLP example. Needs AWS credentials and Nova Pro model access.
+Run: python examples/zero-code-examples/agentcore/run.py
+"""
+
+import os
+import random
+
+from dotenv import load_dotenv
+from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from strands import Agent, tool
+from strands.models import BedrockModel
+from strands.telemetry import StrandsTelemetry
+
+load_dotenv(override=True)
+os.environ.setdefault("OTEL_SEMCONV_STABILITY_OPT_IN", "gen_ai_latest_experimental")
+
+
+@tool
+def roll_die(sides: int = 6) -> dict:
+    """Roll a die with the given number of sides."""
+    result = random.randint(1, sides)
+    return {"sides": sides, "result": result, "message": f"Rolled a {sides}-sided die and got {result}"}
+
+
+@tool
+def check_prime(n: int) -> bool:
+    """Return True if number is prime."""
+    return n >= 2 and all(n % i for i in range(2, int(n**0.5) + 1))
+
+
+def main():
+    if not os.getenv("AWS_DEFAULT_REGION"):
+        print("AWS_DEFAULT_REGION not set.")
+        return
+    os.environ.setdefault("OTEL_RESOURCE_ATTRIBUTES",
+        "agentevals.eval_set_id=agentcore_eval,agentevals.session_name=agentcore-zero-code")
+    telemetry = StrandsTelemetry()
+    telemetry.tracer_provider.add_span_processor(BatchSpanProcessor(OTLPSpanExporter(), schedule_delay_millis=1000))
+    agent = Agent(
+        model=BedrockModel(model_id="us.amazon.nova-pro-v1:0"),
+        tools=[roll_die, check_prime],
+        system_prompt="You are a helpful assistant. Use roll_die when asked to roll dice. Use check_prime when asked about prime numbers.",
+        name="dice_agent",
+    )
+    queries = ["Hi! Can you help me?", "Roll a 20-sided die for me", "Is the number you rolled prime?"]
+    for i, query in enumerate(queries, 1):
+        print(f"\n[{i}/{len(queries)}] User: {query}\n     Agent: {agent(query)}")
+    telemetry.tracer_provider.force_flush()
+    print("All traces flushed.")
+
+
+if __name__ == "__main__":
+    main()