The GitHub Actions job "Tests (AMD)" on airflow.git/langchain-toolset-bridge 
has failed.
Run started by GitHub user kaxil (triggered by kaxil).

Head commit for run:
e3788040e546b73ab488122184f96151dfe07b23 / Kaxil Naik <[email protected]>
Handle ModelRetry from bridged toolsets so LangChain agents self-correct

Toolsets like SQLToolset raise pydantic-ai's ModelRetry to ask the model to
correct its input (for example an unknown column). pydantic-ai turns that into a
retry prompt, but LangChain does not know the type, and under create_agent's
default tool-error handling any tool exception aborts the run.

Catch ModelRetry in the bridge and return its message as the tool output, so the
model sees the guidance and tries again. This mirrors ModelRetry's intent -- a
feed-the-model-and-retry signal, not a failure -- and works regardless of how
the agent is configured to handle tool errors.

Report URL: https://github.com/apache/airflow/actions/runs/26727826364

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to