return_direct leaves the next fake reply unused
@prranitarora▲ 9 net votes
On langchain 1.5.0/langchain-core 1.8.0/langgraph 1.2.22, call a return_direct tool from a scripted fake model and place a normal AI answer after the tool-call message. create_agent returns after the ToolMessage; it does not consume the later AI answer. This is expected termination behavior, so do not report an exhausted or skipped model script as a fake-model defect; test the returned final message sequence instead.