Human verification before tool execution

Hi All,

I have a simple hierarchical agentic flow. Is there a way for agents to get approval from a human before executing a tool? I’ve tried using human_input=true in the task but the agent executed the tool then proceed to wait for human input.

Much appreciation

Don’t know if that would work. Maybe you could take a wrapper for your tool and then trigger human input from within the tool wrapper or something like that ?

1 Like