JoshuaC215
tool artifact support
Hi, I'm curious if you have plans to support tools when the response format is content_and_artifact.
This will help out a lot when the tool output is very long and the agent does not need it in the context window. For example, it can be an entire table for a database or a query result with lots of rows.
Since tools in langchain does not support updating an agent state, I guess this will temporarily be the only way for handling long tool outputs.
https://python.langchain.com/docs/how_to/tool_artifacts/
Hey, I think the repo is flexible to support it. I quickly looked and didn't see an example of existing langgraph code that uses the artifacts. If you can find or generate a working example for this, we could see about whether any changes are needed in the core repo to better demonstrate or support it.
Langgraph supports updating agent state from tools: https://langchain-ai.github.io/langgraph/how-tos/update-state-from-tools/
