Prerequisites
What are you trying to do that currently feels hard or impossible?
Right now BigQuery tools based on SQL query, such as bigqueryexecutesql, bigquerysql, don't have protection against large number of results returned by a SQL query. This could result in:
- The tool taking really long time to fetch the entire result which may not be useful (think conversational user journeys), or even timing out
- A large size output consuming a lot of LLM tokens
Ideally it should be made configurable for the agent developer so that they can protect their agent from such scenarios.
For a comparison, ADK BQ tools have this configurable: https://github.com/google/adk-python/blob/30d3411d603f12ca5bcdd2d71773d087f3191dba/src/google/adk/tools/bigquery/config.py#L73C3-L73C24
Suggested Solution(s)
No response
Alternatives Considered
No response
Additional Details
No response
Prerequisites
What are you trying to do that currently feels hard or impossible?
Right now BigQuery tools based on SQL query, such as
bigqueryexecutesql,bigquerysql, don't have protection against large number of results returned by a SQL query. This could result in:Ideally it should be made configurable for the agent developer so that they can protect their agent from such scenarios.
For a comparison, ADK BQ tools have this configurable: https://github.com/google/adk-python/blob/30d3411d603f12ca5bcdd2d71773d087f3191dba/src/google/adk/tools/bigquery/config.py#L73C3-L73C24
Suggested Solution(s)
No response
Alternatives Considered
No response
Additional Details
No response