diff --git a/src/google/adk/flows/llm_flows/functions.py b/src/google/adk/flows/llm_flows/functions.py
index 664d0d6c67..605d6fe8ba 100644
--- a/src/google/adk/flows/llm_flows/functions.py
+++ b/src/google/adk/flows/llm_flows/functions.py
@@ -21,6 +21,7 @@
 import copy
 import functools
 import inspect
+import json
 import logging
 import threading
 from typing import Any
@@ -962,6 +963,20 @@ def __build_response_event(
       parts=[part_function_response],
   )
 
+  # When summarization is skipped, ensure a displayable text part is added.
+  if tool_context.actions.skip_summarization:
+    # If the tool returned a non-dict, it was wrapped in {'result': ...}.
+    # This unwraps the value for display; otherwise, it uses the original dict.
+    result_payload = function_result.get('result', function_result)
+    if isinstance(result_payload, str):
+      result_text = result_payload
+    else:
+      # Safely serialize non-string results to JSON for display.
+      result_text = json.dumps(
+          result_payload, ensure_ascii=False, default=str
+      )
+    content.parts.append(types.Part.from_text(text=result_text))
+
   function_response_event = Event(
       invocation_id=invocation_context.invocation_id,
       author=invocation_context.agent.name,
diff --git a/tests/unittests/tools/test_agent_tool.py b/tests/unittests/tools/test_agent_tool.py
index b5f59be0fc..49341c07ce 100644
--- a/tests/unittests/tools/test_agent_tool.py
+++ b/tests/unittests/tools/test_agent_tool.py
@@ -14,6 +14,7 @@
 
 from typing import Any
 from typing import Optional
+import pytest
 
 from google.adk.agents.callback_context import CallbackContext
 from google.adk.agents.invocation_context import InvocationContext
@@ -1164,3 +1165,82 @@ def test_empty_sequential_agent_falls_back_to_request(self):
 
     # Should fall back to 'request' parameter
     assert declaration.parameters.properties['request'].type == 'STRING'
+
+
+@pytest.fixture
+def setup_skip_summarization_runner():
+  def _setup_runner(tool_agent_model_responses, tool_agent_output_schema=None):
+    tool_agent_model = testing_utils.MockModel.create(responses=tool_agent_model_responses)
+    tool_agent = Agent(
+        name="tool_agent",
+        model=tool_agent_model,
+        output_schema=tool_agent_output_schema
+    )
+
+    agent_tool = AgentTool(agent=tool_agent, skip_summarization=True)
+
+    root_agent_model = testing_utils.MockModel.create(
+        responses=[
+            function_call_no_schema,
+            "final_summary_text_that_should_not_be_reached",
+        ]
+    )
+
+    root_agent = Agent(
+        name="root_agent",
+        model=root_agent_model,
+        tools=[agent_tool],
+    )
+    return testing_utils.InMemoryRunner(root_agent)
+  return _setup_runner
+
+
+def test_agent_tool_skip_summarization_has_text_output(setup_skip_summarization_runner):
+  """Tests that when skip_summarization is True, the final event contains text content."""
+  runner = setup_skip_summarization_runner(tool_agent_model_responses=["tool_response_text"])
+  events = runner.run("start")
+
+  final_events = [e for e in events if e.is_final_response()]
+  assert final_events
+  last_event = final_events[-1]
+  assert last_event.is_final_response()
+
+  assert any(p.function_response for p in last_event.content.parts)
+
+  assert [p.text for p in last_event.content.parts if p.text] == ["tool_response_text"]
+
+
+def test_agent_tool_skip_summarization_preserves_json_string_output(setup_skip_summarization_runner):
+  """Tests that structured output string is preserved as text when skipping summarization."""
+  runner = setup_skip_summarization_runner(tool_agent_model_responses=['{"field": "value"}'])
+  events = runner.run("start")
+
+  final_events = [e for e in events if e.is_final_response()]
+  assert final_events
+  last_event = final_events[-1]
+  assert last_event.is_final_response()
+
+  text_parts = [p.text for p in last_event.content.parts if p.text]
+  
+  # Check that the JSON string content is preserved exactly
+  assert text_parts == ['{"field": "value"}']
+
+
+def test_agent_tool_skip_summarization_handles_non_string_result(setup_skip_summarization_runner):
+  """Tests that non-string (dict) output is correctly serialized as JSON text."""
+  class CustomOutput(BaseModel):
+      value: int
+
+  runner = setup_skip_summarization_runner(
+      tool_agent_model_responses=['{"value": 123}'],
+      tool_agent_output_schema=CustomOutput
+  )
+  events = runner.run("start")
+
+  final_events = [e for e in events if e.is_final_response()]
+  assert final_events
+  last_event = final_events[-1]
+  
+  text_parts = [p.text for p in last_event.content.parts if p.text]
+  
+  assert text_parts == ['{"value": 123}']