[Enhancement] Able to make agent wait for user input in agent-chaining when mentioned

[ksajan]

Agent-Chaining is awesome when I want to use multiple tools that are interlinked when one output is required for another tool. But in some case it needs to verify the input from the user. In agent multichain it does ask for it but it doesn't wait it just fires the next agent. Is it possible to make it wait?

[Dannypeja]

I also found that the agent chain is too much of a chain. It fires all agents in sequence.
For a true More approach we should be able to have a master agent which determines which next agent should handle the request.
That would be amazing for building complex systems.
Btw one could probably pretty much replicate the ChatGPT experience with that. Use agents as default with less need to switch endpoints