[elmaroun07-hub]

Link to Issue or Description of Change

2. Or, if no issue exists, describe the change:

Problem:
When an API answers 2xx with a body that is empty or not valid JSON, RestApiTool.call returns only {"text": response.text}. With an empty body the model receives {"text": ""} and nothing tells it that the API returned no content, so it cannot distinguish "nothing came back" from data.

Solution:
Keep "text" exactly as it is and add one "message" key saying the body was empty, or was not valid JSON, with the status code. The call still counts as a success, so nothing is reported under "error" and _detect_error_in_response is unaffected. "message" is the key this file already uses for neutral notes ({"pending": True, "message": ...}). JSON responses and error statuses are unchanged.

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally. (See below: the openapi_tool folder passes on four Python versions; the full tox run did not finish on my machine.)

Added to test_rest_api_tool.py: empty body, whitespace-only body, HTML body, plain-text body, and a JSON body returned unchanged. Each test compares the whole returned dict. Before the fix: 4 failed, 1 passed. After: 5 passed.

python -m pytest tests/unittests/tools/openapi_tool, run in each tox environment (Windows):

Python	Result
3.11.17	440 passed
3.12.10	440 passed
3.13.18	440 passed
3.14.10	440 passed

Not completed: the full tox run. All four environments built, but the full suites were still running after 50 minutes on my Windows machine and I stopped them. Three pre-commit hooks (check-new-py-prefix, compliance-checks, check-docs) could not start under pre-commit on Windows; I ran their scripts directly and each exited 0. addlicense was not installed locally. ruff, isort, pyink and codespell passed.

Manual End-to-End (E2E) Tests:

Tool-level test: a local HTTP server on 127.0.0.1 and a direct RestApiTool.call against it (not run through adk web or a Runner).

code under test: upstream/main (098eefa24)
server: 200 text/plain, body=b''
  tool returned: {"text": ""}
server: 200 text/html; charset=utf-8, body=b'<html><body><h1>Saved</h1></body></html>'
  tool returned: {"text": "<html><body><h1>Saved</h1></body></html>"}

code under test: this branch
server: 200 text/plain, body=b''
  tool returned: {"text": "", "message": "The response body was empty. Status Code: 200"}
server: 200 text/html; charset=utf-8, body=b'<html><body><h1>Saved</h1></body></html>'
  tool returned: {"text": "<html><body><h1>Saved</h1></body></html>", "message": "The response body was not valid JSON. Status Code: 200"}

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 or that my feature works.
• New and existing unit tests pass locally with my changes. (The openapi_tool folder only, as above.)
• I have manually tested my changes end-to-end. (Tool-level, as above.)
• Any dependent changes have been merged and published in downstream modules. (None.)

Additional context

If you would rather flag only the empty body and leave non-empty non-JSON bodies as they are, I can narrow the change.

I found this while measuring how agents respond when a tool returns an empty 200 body; in my runs with another framework, models often answered with made-up values instead of reporting it: https://github.com/elmaroun07-hub/reported-or-invented

This change was written with AI assistance (Claude Code); I reviewed and tested it.