fix(tools): return a structured error when RestApiTool is missing a required path param - #7283

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: RestApiTool raises uncaught KeyError when a required path param is omitted #7282

Problem:
FunctionTool returns {"error": ...} when a mandatory argument is missing so the model can see the failure and retry. RestApiTool does not: _prepare_request_params fills path_params only for keys the caller actually supplied, then does self.endpoint.path.format(**path_params). If a required path parameter is omitted (and has no schema default), the format template still has an unfilled {placeholder}, so str.format raises an uncaught KeyError with the OpenAPI parameter's original name (e.g. 'userId'). RestApiTool.call() only catches httpx.TimeoutException / httpx.HTTPStatusError, so this KeyError propagates out of the tool and aborts the whole agent invocation instead of giving the model a retryable error.

Solution:
Wrap the self._prepare_request_params(...) call in RestApiTool.call() in a try/except KeyError, mirroring the structured-error pattern FunctionTool already uses for missing mandatory args. On a missing path parameter, call() now returns {"error": "Tool <name> execution failed. ... Missing required path parameter '<name>'."} instead of raising.

Testing Plan

Unit Tests:

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

Added test_call_missing_required_path_param_returns_error, which builds a RestApiTool for GET /users/{userId}/messages with a required userId path parameter that has no default, calls it with no args, and asserts the tool returns a structured {"error": ...} naming userId instead of raising, and that no HTTP request is made.

Regression proof — reverting only the source fix and running the new test:

$ git checkout HEAD~1 -- src/google/adk/tools/openapi_tool/openapi_spec_parser/rest_api_tool.py
$ pytest tests/unittests/tools/openapi_tool/openapi_spec_parser/test_rest_api_tool.py -k missing_required_path_param -q
...
E     KeyError: 'userId'
src/google/adk/tools/openapi_tool/openapi_spec_parser/rest_api_tool.py:427: KeyError
1 failed, 81 deselected in 1.23s

With the fix restored:

$ pytest tests/unittests/tools/openapi_tool/openapi_spec_parser/test_rest_api_tool.py -k missing_required_path_param -q
1 passed, 81 deselected in 2.77s

$ pytest tests/unittests/tools/openapi_tool/openapi_spec_parser/test_rest_api_tool.py -q
82 passed, 17 warnings in 2.01s

$ pytest tests/unittests/tools/ -q
2470 passed, 808 warnings in 53.18s

Formatting/lint on the changed files:

$ python -m isort --settings-path pyproject.toml --check --diff <files>
(no changes)
$ python -m pyink --config pyproject.toml --check --diff <files>
All done! 2 files would be left unchanged.
$ python -m ruff check --config pyproject.toml <files>
All checks passed!

Checklist

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

AI assistance disclosure

This change was prepared with the assistance of Claude Code (Anthropic), under human review before submission.

🤖 Generated with Claude Code