### Answer
To prevent interruptions during tool calls that cause hallucinations, enable
Block Interruptions on your conversation node [(1) ](
) . When enabled, the agent will not be interrupted by the user when speaking .
Additionally, based on community experience, if you're in the middle of a tool call and the user interrupts, you should complete the function call and not interrupt the agent [(2) ](
) . You can provide an intermediate wait message in that case .
For your specific flow (conversation node → function node with success/fail transitions → conversation node) , the hallucination issue occurs because the agent lacks proper context when interrupted mid-function-call. The recommended approach is:
1. Enable
Block Interruptions on the conversation node before the function call [(1) ](
)
2. Ensure your function node completes its execution before allowing transitions [(2) ](
)
3. If hallucinations persist after function calls, check that your transition conditions are clear and that the agent is properly grounded to the tool result [(3) ](
)
If the agent continues hallucinating after successful function execution, verify that your conversation nodes following the function call have clear instructions that reference the function's output [(4) ](
) .