Hi Crew AI Community,
We have a custom tool that performs specific tasks and raises errors when issues occur. However, when an error is encountered, the output displays a plain-text summary of the error caused by the internal behavior of Crew, rather than the actual error thrown by the tool.
How can we modify this behavior so that the output provides the “actual error thrown by the tool” instead of a summarized version? Additionally, if the tool encounters an error during a specific task, it should halt execution rather than proceeding to the next task.