Agent Instructions Help
I am struggling with recreating an existing agent in the new orchestration. The original agent worked perfectly, but I am now rebuilding it using the new orchestration and cannot get it to behave in the same way.
The agent is designed to answer questions using knowledge sources stored on SharePoint. It answers the questions accurately, but it always seems to ask follow-up or leading questions afterwards, such as: “Would you like me to provide a general list of common steps?”
What I want is for the agent to only answer the question that has been asked and then end the conversation. It should not ask follow-up questions, offer additional information, or suggest what it could do next.
With the old orchestration, I found that adding strict rules to the instructions was enough to achieve this behaviour. However, I have recreated the agent using the new orchestration and, even with the same strict instructions, it continues to ask follow-up questions and offer additional help.
Is there a different configuration or approach in the new orchestration that I need to use to replicate the behaviour of the original agent?