fix(agent): classify think-only empty responses before retrying

tests/test_run_agent.py

@@ -170,13 +170,21 @@ def _mock_tool_call(name="web_search", arguments="{}", call_id=None):
 
 
 def _mock_response(
-    content="Hello", finish_reason="stop", tool_calls=None, reasoning=None, usage=None
+    content="Hello",
+    finish_reason="stop",
+    tool_calls=None,
+    reasoning=None,
+    reasoning_content=None,
+    reasoning_details=None,
+    usage=None,
 ):
     """Return a SimpleNamespace mimicking an OpenAI ChatCompletion response."""
     msg = _mock_assistant_msg(
         content=content,
         tool_calls=tool_calls,
         reasoning=reasoning,
+        reasoning_content=reasoning_content,
+        reasoning_details=reasoning_details,
     )
     choice = SimpleNamespace(message=msg, finish_reason=finish_reason)
     resp = SimpleNamespace(choices=[choice], model="test/model")
@@ -1498,6 +1506,75 @@ class TestRunConversation:
         assert result["completed"] is True
         assert result["final_response"] == "internal reasoning"
 
+    def test_empty_content_local_resumed_session_triggers_compression(self, agent):
+        """Local resumed reasoning-only responses should compress before burning retries."""
+        self._setup_agent(agent)
+        agent.base_url = "http://127.0.0.1:1234/v1"
+        agent.compression_enabled = True
+        empty_resp = _mock_response(
+            content=None,
+            finish_reason="stop",
+            reasoning_content="reasoning only",
+        )
+        ok_resp = _mock_response(content="Recovered after compression", finish_reason="stop")
+        prefill = [
+            {"role": "user", "content": "old question"},
+            {"role": "assistant", "content": "old answer"},
+        ]
+
+        with (
+            patch.object(agent, "_interruptible_api_call", side_effect=[empty_resp, ok_resp]),
+            patch.object(agent, "_compress_context") as mock_compress,
+            patch.object(agent, "_persist_session"),
+            patch.object(agent, "_save_trajectory"),
+            patch.object(agent, "_cleanup_task_resources"),
+        ):
+            mock_compress.return_value = (
+                [{"role": "user", "content": "compressed user message"}],
+                "compressed system prompt",
+            )
+            result = agent.run_conversation("hello", conversation_history=prefill)
+
+        mock_compress.assert_called_once()
+        assert result["completed"] is True
+        assert result["final_response"] == "Recovered after compression"
+        assert result["api_calls"] == 1  # compression retry is refunded, same as explicit overflow path
+
+    def test_empty_content_repeated_structured_reasoning_salvages_early(self, agent):
+        """Repeated identical structured reasoning-only responses should stop retrying early."""
+        self._setup_agent(agent)
+        empty_resp = _mock_response(
+            content=None,
+            finish_reason="stop",
+            reasoning_content="structured reasoning answer",
+        )
+        agent.client.chat.completions.create.side_effect = [empty_resp, empty_resp]
+        with (
+            patch.object(agent, "_persist_session"),
+            patch.object(agent, "_save_trajectory"),
+            patch.object(agent, "_cleanup_task_resources"),
+        ):
+            result = agent.run_conversation("answer me")
+        assert result["completed"] is True
+        assert result["final_response"] == "structured reasoning answer"
+        assert result["api_calls"] == 2
+
+    def test_empty_content_local_custom_error_is_actionable(self, agent):
+        """Local/custom retries should return a diagnostic tailored to context/endpoint mismatch."""
+        self._setup_agent(agent)
+        agent.base_url = "http://127.0.0.1:1234/v1"
+        empty_resp = _mock_response(content=None, finish_reason="stop")
+        agent.client.chat.completions.create.side_effect = [empty_resp, empty_resp, empty_resp]
+        with (
+            patch.object(agent, "_persist_session"),
+            patch.object(agent, "_save_trajectory"),
+            patch.object(agent, "_cleanup_task_resources"),
+        ):
+            result = agent.run_conversation("answer me")
+        assert result["completed"] is False
+        assert "Local/custom backend returned reasoning-only output" in result["error"]
+        assert "wrong /v1 endpoint" in result["error"]
+
     def test_nous_401_refreshes_after_remint_and_retries(self, agent):
         self._setup_agent(agent)
         agent.provider = "nous"