GitHub user joeyutong added a comment to the discussion: 
[Discussion][Observability 2/2] Built-in Operational Metrics for Flink Agents

You are right: Python currently has no `ToolResponse` equivalent. A normal 
Python `Tool.call` return is treated as success; only resource lookup, argument 
preparation, or invocation exceptions are failures. Java additionally treats an 
unsuccessful `ToolResponse` as failed. I corrected sections 5.4 and 7, as well 
as the documentation in #955, to describe this existing API difference instead 
of claiming an explicit Python error-result normalization. Metrics still 
consume the common `success` / `failed` execution lifecycle outcomes.

GitHub link: 
https://github.com/apache/flink-agents/discussions/901#discussioncomment-17880274

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]

Reply via email to