Confused about the meaning of Graph edges

[sergio-onemedical]

Hi there,

The documentation states that "Edges define dependencies and information flow between nodes". As such, I was expecting that a node would execute only when all of its dependencies are met, but that's not the behavior that's implemented.

Graph._is_node_ready_with_conditions returns True when any of the node dependencies are met rather than when all dependencies are met.

Here's a quick example that demonstrates what I'd consider to be unintuitive behavior. When the code runs, agent3 doesn't see the output of agent2, and doesn't know what to do.

agent1 = Agent(name="agent1", system_prompt="Output the letter 'X' and absolutely nothing else.")
agent2 = Agent(name="agent2", system_prompt="Output a word that starts with the letter you received from agent1, and absolutely nothing else.")
agent3 = Agent(name="agent3", system_prompt="You concatenate the outputs from agent1 and agent2. Do not output anything else.")

builder = GraphBuilder()

builder.add_node(agent1, "agent1")
builder.add_node(agent2, "agent2")
builder.add_node(agent3, "agent3")

builder.add_edge("agent1", "agent2")
builder.add_edge("agent2", "agent3")  # <-- THIS LINE DOESN'T DO ANYTHING
builder.add_edge("agent1", "agent3")

builder.set_entry_point("agent1")

graph = builder.build()

result = graph("Follow your system prompt")

print(agent3.messages)

Could someone help me clarify whether this is a bug or a misunderstanding on my part about what edges are supposed to mean?
Thanks!