Learning from experience of building the Graph Inspector API, I am realizing that there's some powerful use cases that we can unlock if we have a node that allows us to inspect what's wired into (or out of) a subgraph that contains this node.
Sample use case: a Router worker. Imagine a new worker that is part of the Agent Kit, which can examine what's wired out of it. This would allow the user of the Router to connect it to multiple destination workers (Writer, Critic, Librarian, etc.) and let the Router decide which destination to choose based on the task given.
I would like to figure out a Core Kit node that enables implementing such a worker.
It seems such a node would need to have a capability to look outside of its own graph, to the enclosed graph, and return back the input/output schemas representing the wiring.
Learning from experience of building the Graph Inspector API, I am realizing that there's some powerful use cases that we can unlock if we have a node that allows us to inspect what's wired into (or out of) a subgraph that contains this node.
Sample use case: a Router worker. Imagine a new worker that is part of the Agent Kit, which can examine what's wired out of it. This would allow the user of the Router to connect it to multiple destination workers (Writer, Critic, Librarian, etc.) and let the Router decide which destination to choose based on the task given.
I would like to figure out a Core Kit node that enables implementing such a worker.
It seems such a node would need to have a capability to look outside of its own graph, to the enclosed graph, and return back the input/output schemas representing the wiring.