Simple agent stop criteria requires no tool calls AND output message item to be present

resources_servers/workbench/configs/workbench.yaml

@@ -24,4 +24,4 @@ workbench_simple_agent:
         - name: example
           type: example
           jsonl_fpath: resources_servers/workbench/data/example.jsonl
-      max_turns: 6
+      max_steps: 6

responses_api_agents/simple_agent/app.py

@@ -33,13 +33,14 @@
     NeMoGymResponse,
     NeMoGymResponseCreateParamsNonStreaming,
     NeMoGymResponseFunctionToolCall,
+    NeMoGymResponseOutputMessage,
 )
 
 
 class SimpleAgentConfig(BaseResponsesAPIAgentConfig):
     resources_server: ResourcesServerRef
     model_server: ModelServerRef
-    max_turns: int = None
+    max_steps: int = None
 
 
 class SimpleAgentRunRequest(BaseRunRequest):
@@ -64,10 +65,10 @@ async def responses(self, body: NeMoGymResponseCreateParamsNonStreaming = Body()
             body.input = [NeMoGymEasyInputMessage(role="user", content=body.input)]
 
         new_outputs = []
-        turn = 0
+        step = 0
 
         while True:
-            turn += 1
+            step += 1
             new_body = body.model_copy(update={"input": body.input + new_outputs})
 
             model_response = await self.server_client.post(
@@ -87,7 +88,10 @@ async def responses(self, body: NeMoGymResponseCreateParamsNonStreaming = Body()
             new_outputs.extend(output)
 
             all_fn_calls: List[NeMoGymResponseFunctionToolCall] = [o for o in output if o.type == "function_call"]
-            if not all_fn_calls:
+            all_output_messages: List[NeMoGymResponseOutputMessage] = [
+                o for o in output if o.type == "message" and o.role == "assistant"
+            ]
+            if not all_fn_calls and all_output_messages:
                 break
 
             for output_function_call in all_fn_calls:
@@ -104,8 +108,8 @@ async def responses(self, body: NeMoGymResponseCreateParamsNonStreaming = Body()
                 )
                 new_outputs.append(tool_response)
 
-            # Check if max turns is not None and if we have exhausted it.
-            if self.config.max_turns and turn >= self.config.max_turns:
+            # Check if max steps is not None and if we have exhausted it.
+            if self.config.max_steps and step >= self.config.max_steps:
                 break
 
         model_response.output = new_outputs

responses_api_agents/simple_agent/tests/test_app.py

@@ -11,14 +11,16 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from unittest.mock import MagicMock
+from unittest.mock import MagicMock, call
 
 from fastapi.testclient import TestClient
 from pytest import MonkeyPatch
 
 from nemo_gym.openai_utils import (
     NeMoGymEasyInputMessage,
     NeMoGymResponseCreateParamsNonStreaming,
+    NeMoGymResponseReasoningItem,
+    NeMoGymSummary,
 )
 from nemo_gym.server_utils import ServerClient
 from responses_api_agents.simple_agent.app import (
@@ -150,3 +152,165 @@ async def test_responses(self, monkeypatch: MonkeyPatch) -> None:
             "user": None,
         }
         assert expected_responses_dict == actual_responses_dict
+
+    async def test_responses_continues_on_reasoning_only(self, monkeypatch: MonkeyPatch) -> None:
+        config = SimpleAgentConfig(
+            host="0.0.0.0",
+            port=8080,
+            entrypoint="",
+            model_server=ModelServerRef(
+                type="responses_api_models",
+                name="my server name",
+            ),
+            resources_server=ResourcesServerRef(
+                type="resources_servers",
+                name="",
+            ),
+        )
+        server = SimpleAgent(config=config, server_client=MagicMock(spec=ServerClient))
+        app = server.setup_webserver()
+        client = TestClient(app)
+
+        mock_response_reasoning_data = {
+            "id": "resp_688babb004988199b26c5250ba69c1e80abdf302bcd600d3",
+            "created_at": 1753983920.0,
+            "model": "dummy_model",
+            "object": "response",
+            "output": [
+                {
+                    "id": "msg_688babb17a7881998cc7a42d53c8e5790abdf302bcd600d3",
+                    "summary": [
+                        {
+                            "text": "I'm thinking how to respond",
+                            "type": "summary_text",
+                        }
+                    ],
+                    "status": "completed",
+                    "type": "reasoning",
+                }
+            ],
+            "parallel_tool_calls": True,
+            "tool_choice": "auto",
+            "tools": [],
+        }
+
+        mock_response_chat_data = {
+            "id": "resp_688babb004988199b26c5250ba69c1e80abdf302bcd600d3",
+            "created_at": 1753983920.0,
+            "model": "dummy_model",
+            "object": "response",
+            "output": [
+                {
+                    "id": "msg_688babb17a7881998cc7a42d53c8e5790abdf302bcd600d3",
+                    "content": [
+                        {
+                            "annotations": [],
+                            "text": "Hello! How can I help you today?",
+                            "type": "output_text",
+                        }
+                    ],
+                    "role": "assistant",
+                    "status": "completed",
+                    "type": "message",
+                }
+            ],
+            "parallel_tool_calls": True,
+            "tool_choice": "auto",
+            "tools": [],
+        }
+
+        dotjson_mock = MagicMock()
+        dotjson_mock.json.side_effect = [mock_response_reasoning_data, mock_response_chat_data]
+        server.server_client.post.return_value = dotjson_mock
+
+        # No model provided should use the one from the config
+        res_no_model = client.post("/v1/responses", json={"input": [{"role": "user", "content": "hello"}]})
+        assert res_no_model.status_code == 200
+
+        expected_calls = [
+            call(
+                server_name="my server name",
+                url_path="/v1/responses",
+                json=NeMoGymResponseCreateParamsNonStreaming(
+                    input=[NeMoGymEasyInputMessage(content="hello", role="user", type="message")]
+                ),
+            ),
+            call().json(),
+            call(
+                server_name="my server name",
+                url_path="/v1/responses",
+                json=NeMoGymResponseCreateParamsNonStreaming(
+                    input=[
+                        NeMoGymEasyInputMessage(content="hello", role="user", type="message"),
+                        NeMoGymResponseReasoningItem(
+                            id="msg_688babb17a7881998cc7a42d53c8e5790abdf302bcd600d3",
+                            summary=[NeMoGymSummary(text="I'm thinking how to respond", type="summary_text")],
+                            type="reasoning",
+                            encrypted_content=None,
+                            status="completed",
+                        ),
+                    ]
+                ),
+            ),
+            call().json(),
+        ]
+        server.server_client.post.assert_has_calls(expected_calls)
+
+        actual_responses_dict = res_no_model.json()
+        expected_responses_dict = {
+            "id": "resp_688babb004988199b26c5250ba69c1e80abdf302bcd600d3",
+            "created_at": 1753983920.0,
+            "error": None,
+            "incomplete_details": None,
+            "instructions": None,
+            "metadata": None,
+            "model": "dummy_model",
+            "object": "response",
+            "output": [
+                {
+                    "id": "msg_688babb17a7881998cc7a42d53c8e5790abdf302bcd600d3",
+                    "encrypted_content": None,
+                    "summary": [
+                        {
+                            "text": "I'm thinking how to respond",
+                            "type": "summary_text",
+                        }
+                    ],
+                    "status": "completed",
+                    "type": "reasoning",
+                },
+                {
+                    "id": "msg_688babb17a7881998cc7a42d53c8e5790abdf302bcd600d3",
+                    "content": [
+                        {
+                            "annotations": [],
+                            "text": "Hello! How can I help you today?",
+                            "type": "output_text",
+                            "logprobs": None,
+                        }
+                    ],
+                    "role": "assistant",
+                    "status": "completed",
+                    "type": "message",
+                },
+            ],
+            "parallel_tool_calls": True,
+            "temperature": None,
+            "tool_choice": "auto",
+            "tools": [],
+            "top_p": None,
+            "background": None,
+            "max_output_tokens": None,
+            "max_tool_calls": None,
+            "previous_response_id": None,
+            "prompt": None,
+            "reasoning": None,
+            "service_tier": None,
+            "status": None,
+            "text": None,
+            "top_logprobs": None,
+            "truncation": None,
+            "usage": None,
+            "user": None,
+        }
+        assert expected_responses_dict == actual_responses_dict

responses_api_agents/simple_agent_stateful/app.py

@@ -33,6 +33,7 @@
     NeMoGymResponse,
     NeMoGymResponseCreateParamsNonStreaming,
     NeMoGymResponseFunctionToolCall,
+    NeMoGymResponseOutputMessage,
 )
 
 
@@ -55,6 +56,7 @@ class SimpleAgentStatefulVerifyResponse(BaseVerifyResponse):
     model_config = ConfigDict(extra="allow")
 
 
+# TODO we should find a way to merge this with the regular simple agent.
 class SimpleAgentStateful(SimpleResponsesAPIAgent):
     config: SimpleAgentStatefulConfig
 
@@ -87,7 +89,10 @@ async def responses(self, body: NeMoGymResponseCreateParamsNonStreaming = Body()
             new_outputs.extend(output)
 
             all_fn_calls: List[NeMoGymResponseFunctionToolCall] = [o for o in output if o.type == "function_call"]
-            if not all_fn_calls:
+            all_output_messages: List[NeMoGymResponseOutputMessage] = [
+                o for o in output if o.type == "message" and o.role == "assistant"
+            ]
+            if not all_fn_calls and all_output_messages:
                 break
 
             for output_function_call in all_fn_calls: