langgraph/how-tos/managing-conversation-history/

[giscus[bot]]

langgraph/how-tos/managing-conversation-history/

Build language agents as graphs

https://langchain-ai.github.io/langgraph/how-tos/managing-conversation-history/

[Erybor]

I'm confused about how the checkpointer and the memory work. How does it know which keys to checkpoint and feed into the model?

In this example only a single key "messages" is defined in the State, what if we have 10 keys but only want to pass in 1 or 2 keys from the Memory to the model?

Does the whole state get passed to the model or is it just "messages"?

Where can I see what actually gets fed into the model prompt from the Memory?