langchain
langchain copied to clipboard
Fix: Add run_manager on all AgentFinish returns in AgentExecutor
Fix: Adds run_manager to all return calls on AgentExecutor
Currently, run_manager
is only passed to _return / _areturn if we stop the execution due to timeout/max-iterations. We also would like to have the callback signal for on_agent_finish
when the agent finished due to tool-return or simply decides to take that action.
Before submitting
Who can review?
Community members can review the PR once tests pass. Tag maintainers/contributors who might be interested:
Project lead @hwchase17
Tracing / Callbacks / Async @agola11