feat(app): return S3 result location in rollout response

Author: luyuzhe111

src/agentcore_rl_toolkit/app.py

@@ -117,7 +117,9 @@ def _validate_and_normalize_rollout(self, rollout_dict: dict) -> dict:
         rollout_dict["rewards"] = rewards
         return rollout_dict
 
-    def save_rollout_and_notify(self, rollout_data: dict, training_config: dict):
+    def save_rollout_and_notify(
+        self, rollout_data: dict, training_config: dict, payload: dict = None, result_key: str = None
+    ):
         """
         Save rollout data to S3 and notify SQS queue.
 
@@ -129,6 +131,8 @@ def save_rollout_and_notify(self, rollout_data: dict, training_config: dict):
                 - exp_id: Experiment ID for organizing data
                 - session_id: Session id for the current task
                 - input_id: id for discriminating different input data examples
+            payload: Original request payload (included in saved result for debugging)
+            result_key: S3 key for the result (computed externally for consistency)
         """
         # Validate and extract training configuration
         try:
@@ -137,17 +141,24 @@ def save_rollout_and_notify(self, rollout_data: dict, training_config: dict):
             logging.error(f"Invalid training configuration: {e}")
             raise
 
-        result_key = f"{config.exp_id}/{config.input_id}_{config.session_id}.json"
+        # Use provided result_key or compute it
+        if result_key is None:
+            result_key = f"{config.exp_id}/{config.input_id}_{config.session_id}.json"
 
         if "status_code" not in rollout_data:
             rollout_data["status_code"] = 200
 
         if "stop_reason" not in rollout_data:
             rollout_data["stop_reason"] = "end_turn"
 
-        # Return the input id identifying rollouts of the same input data (prompt) example
-        # for advantage computation.
+        # Include metadata for correlation and debugging
         rollout_data["input_id"] = config.input_id
+        rollout_data["s3_bucket"] = config.s3_bucket
+        rollout_data["result_key"] = result_key
+
+        # Include full payload for debugging (with _training config for reproducibility)
+        if payload is not None:
+            rollout_data["payload"] = payload
 
         # Save to S3
         try:
@@ -205,7 +216,7 @@ def invoke_agent(payload, context):  # Can be sync or async
             Decorated function registered as entrypoint
         """
 
-        async def rollout_background_task(payload, context):
+        async def rollout_background_task(payload, context, result_key):
             """Background task that does the actual agent work and rollout saving."""
             training_config = payload.get("_training")
 
@@ -225,7 +236,12 @@ async def rollout_background_task(payload, context):
 
                 # Save rollout data if we have training config
                 if isinstance(result, dict) and training_config:
-                    self.save_rollout_and_notify(rollout_data=result, training_config=training_config)
+                    self.save_rollout_and_notify(
+                        rollout_data=result,
+                        training_config=training_config,
+                        payload=payload,
+                        result_key=result_key,
+                    )
                     logging.info(f"Rollout data saved for function: {func.__name__}")
 
                 return result
@@ -234,7 +250,12 @@ async def rollout_background_task(payload, context):
                 # Always save error rollout for client notification
                 if training_config:
                     error_rollout = {"status_code": 500, "stop_reason": str(e)}
-                    self.save_rollout_and_notify(rollout_data=error_rollout, training_config=training_config)
+                    self.save_rollout_and_notify(
+                        rollout_data=error_rollout,
+                        training_config=training_config,
+                        payload=payload,
+                        result_key=result_key,
+                    )
                     logging.error(f"Error rollout saved for function: {func.__name__}: {e}")
                 raise
             finally:
@@ -244,8 +265,26 @@ async def rollout_background_task(payload, context):
         @wraps(func)
         async def rollout_entrypoint_wrapper(payload, context):
             """Entrypoint that starts background task and returns immediately."""
+            training_config = payload.get("_training")
+
+            # Compute result_key upfront so we can return it to the client
+            result_key = None
+            if training_config:
+                exp_id = training_config.get("exp_id", "")
+                input_id = training_config.get("input_id", "")
+                session_id = training_config.get("session_id", "")
+                result_key = f"{exp_id}/{input_id}_{session_id}.json"
+
             # Start background task without waiting
-            asyncio.create_task(rollout_background_task(payload, context))
+            asyncio.create_task(rollout_background_task(payload, context, result_key))
+
+            # Return result location so client can poll S3 for completion
+            if training_config:
+                return {
+                    "status": "processing",
+                    "s3_bucket": training_config.get("s3_bucket"),
+                    "result_key": result_key,
+                }
             return {"status": "processing"}
 
         # Remove __wrapped__ so inspect.signature() sees the wrapper's actual signature

tests/test_rollout_entrypoint.py

@@ -82,3 +82,51 @@ def handler(payload: dict):
 
     assert response.status_code == 200
     assert response.json() == {"status": "processing"}
+
+
+def test_response_includes_result_location_with_training_config():
+    """Test that response includes s3_bucket and result_key when _training config is provided."""
+    app = AgentCoreRLApp()
+
+    @app.rollout_entrypoint
+    async def handler(payload: dict):
+        return {"rollout_data": [{"test": True}], "rewards": [1.0]}
+
+    client = TestClient(app)
+    response = client.post(
+        "/invocations",
+        json={
+            "prompt": "test",
+            "_training": {
+                "exp_id": "exp-123",
+                "session_id": "sess-456",
+                "input_id": "input-789",
+                "s3_bucket": "my-bucket",
+                "sqs_url": "https://sqs.us-east-1.amazonaws.com/123/queue",
+            },
+        },
+    )
+
+    assert response.status_code == 200
+    result = response.json()
+    assert result["status"] == "processing"
+    assert result["s3_bucket"] == "my-bucket"
+    assert result["result_key"] == "exp-123/input-789_sess-456.json"
+
+
+def test_response_without_training_config():
+    """Test that response is minimal when no _training config is provided."""
+    app = AgentCoreRLApp()
+
+    @app.rollout_entrypoint
+    async def handler(payload: dict):
+        return {"rollout_data": [{"test": True}], "rewards": [1.0]}
+
+    client = TestClient(app)
+    response = client.post("/invocations", json={"prompt": "test"})
+
+    assert response.status_code == 200
+    result = response.json()
+    assert result == {"status": "processing"}
+    assert "s3_bucket" not in result
+    assert "result_key" not in result