fix(agent_tool): ensure text output when skip_summarization is True

[ItsMacto]

When skip_summarization=True, AgentTool previously terminated the flow with a FunctionResponse event that contained no text, leaving UIs stuck.

This change appends a Text part to the response event when summarization is skipped, so the tool's output is displayed. The implementation avoids duplication, safely serializes non-string results, and preserves existing behavior for normal flows.

Added regression tests to verify text output is present for both string and JSON-like tool results.

Please ensure you have read the contribution guide before creating a pull request.

Link to Issue or Description of Change

1. Link to an existing issue (if applicable):

• Closes: The agent doesn't work as expected when "skip_summarization" is set to True in AgentTool #3881

Problem:
When skip_summarization=True, the final response event emitted by AgentTool contains only a FunctionResponse and no displayable text. Most UIs do not render function responses, making the agent appear to hang or return an empty result.

Solution:
When skip summarization is enabled, the function result is additionally included as a Text part in the response event. If the result is a string, it is used directly; otherwise, it is safely serialized to JSON for display. This ensures the final event remains displayable while preserving existing behavior for non-skip paths.

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally.

Passed tests:

• pytest tests/unittests/tools/test_agent_tool.py

Manual End-to-End (E2E) Tests:

Not required. The change affects event construction and is fully covered by unit tests validating the final response content.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have added tests that prove my fix is effective.
• New and existing unit tests pass locally with my changes.
• I have manually tested my changes end-to-end.
• Any dependent changes have been merged and published in downstream modules.

[gemini-code-assist]

Summary of Changes

Hello @ItsMacto, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a critical user experience issue where the AgentTool would not provide any visible output to the user interface when the skip_summarization flag was enabled. The change ensures that the raw output of the tool, whether a simple string or a complex JSON structure, is always included as a displayable text part in the final response event. This prevents UIs from appearing to hang and provides immediate feedback to the user, significantly improving the interaction flow.

Highlights

• Fix for UI hang with skip_summarization: When skip_summarization=True in AgentTool, the system previously terminated the flow with a FunctionResponse event that lacked displayable text, causing UIs to appear stuck. This PR resolves that by ensuring a text output is always present.
• Ensuring displayable text output: A Text part is now appended to the response event when summarization is skipped. If the tool's result is a string, it's used directly; otherwise, non-string results are safely serialized to JSON for display.
• Regression tests added: New regression tests have been introduced to verify that text output is correctly present for both string and JSON-like tool results when summarization is skipped, ensuring the fix is robust.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

• src/google/adk/flows/llm_flows/functions.py
  • Imported the json module to handle serialization of non-string results.
  • Added conditional logic within __build_response_event to append a Text part to the event's content if tool_context.actions.skip_summarization is true. This text part contains the tool's result, serialized to JSON if not already a string.
• tests/unittests/tools/test_agent_tool.py
  • Added test_agent_tool_skip_summarization_has_text_output to verify that a final event contains text content when skip_summarization is enabled and the tool returns a string.
  • Added test_agent_tool_skip_summarization_preserves_json_string_output to confirm that structured JSON output is correctly preserved and included as text when skip_summarization is enabled.

Activity

• No human activity (comments, reviews) has been recorded on this pull request yet.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[ItsMacto]

/gemini review

[gemini-code-assist]

Code Review

This pull request addresses an issue where UIs could get stuck when skip_summarization=True by ensuring a text part is always included in the final event. The implementation correctly adds a text representation of the tool's output, and includes new regression tests.

My review includes a suggestion to refine the serialization logic for non-string results to provide a cleaner output. Additionally, I've recommended strengthening the new test assertions and expanding test coverage to better validate the fix.

[gemini-code-assist]

Code Review

This pull request addresses an issue where AgentTool would not produce a displayable text output when skip_summarization is enabled, causing UIs to appear stuck. The fix correctly appends a Text part to the response event, serializing non-string results to ensure there's always a visible output. The added regression tests effectively cover the new functionality. My review includes suggestions to strengthen the assertions in these new tests to make them more precise and less prone to false positives.

[ItsMacto]

/gemini review

[gemini-code-assist]

Code Review

This pull request fixes an issue where the UI would hang when skip_summarization=True by ensuring a text part is always included in the final event. The implementation correctly handles both string and non-string results from the tool. The accompanying tests are thorough and cover the key scenarios. My review includes a suggestion to refactor the implementation for better readability and another to reduce code duplication in the new tests using a pytest fixture.

[ItsMacto]

/gemini review

[gemini-code-assist]

Code Review

This pull request effectively addresses the issue of UIs hanging when skip_summarization=True by ensuring a text part is always included in the function response event. The implementation correctly handles both string and non-string (JSON-serializable) tool outputs. The new unit tests are comprehensive and validate the fix across different scenarios. The changes are well-implemented and solve the problem described. I have one minor suggestion to improve code clarity and maintainability.

[ryanaiagent]

Hi @ItsMacto , Thank you for your contribution! We appreciate you taking the time to submit this pull request. Can you please fix the formatting errors. You can run autoformat.sh. Also can you also fix the failing unit tests.

[ItsMacto]

/gemini review

[gemini-code-assist]

Code Review

This pull request fixes an issue where the UI would hang when skip_summarization=True by ensuring a text part is always included in the final event. The implementation correctly handles both string and non-string tool outputs, and includes a good set of regression tests to prevent this issue from recurring.

My review includes one suggestion to improve the robustness of the logic that unwraps tool results. The current implementation could lead to data loss if a tool returns a dictionary that happens to contain a result key. The suggested change makes this logic more precise to avoid this potential bug.

Overall, this is a solid fix for an important usability issue.

[ItsMacto]

/gemini review

[gemini-code-assist]

Code Review

This pull request effectively resolves an issue where the UI would hang when skip_summarization=True by ensuring a text part is always included in the final response event. The implementation is robust, correctly handling different types of tool results (strings, JSON, and other objects) and safely serializing them. The logic also correctly avoids adding a text part for error responses. The addition of comprehensive regression tests covering various output scenarios is excellent and ensures the fix is effective and won't be broken by future changes. Overall, this is a high-quality contribution that improves the user experience and stability of the agent tool.

[ItsMacto]

Hi @ryanaiagent , I believe I made the changes correctly. One thing when running autoformat.sh. it affects like 50 files that are not mine. I just didn't commit these changes, but I'm not sure what the best course of action is.