NeMo Agent Toolkit with Langgraph In Memory Saver

AI & Data Science NVIDIA NeMo
1

How can I use NeMo Agent Toolkit with Langgraph Checkpointer Memory saver that required thread_id to invoke graph ?

2

Hi there,

You can wrap your LangGraph agent using the NeMo langgraph_wrapper and handle thread_id inside your graph’s entry function (where you call graph.invoke(…, config={“thread_id”: …})). For details and a working example, see:

NeMo “Running Existing LangGraph Agents” docs: Running Existing LangGraph Agents in NVIDIA NeMo Agent Toolkit — NVIDIA NeMo Agent Toolkit (1.8)

“Running Existing Agents in NeMo Agent Toolkit” overview: Running Existing Agents in NVIDIA NeMo Agent Toolkit — NVIDIA NeMo Agent Toolkit (1.8)

NeMo LangGraph deep research example (shows how a LangGraph agent is wired into NeMo): NeMo-Agent-Toolkit/examples/frameworks/auto_wrapper/langchain_deep_research at develop · NVIDIA/NeMo-Agent-Toolkit · GitHub
github

NeMo Agent Toolkit GitHub repo (for latest examples and issues): GitHub - NVIDIA/NeMo-Agent-Toolkit: The NVIDIA NeMo Agent toolkit is an open-source library for efficiently connecting and optimizing teams of AI agents. · GitHub

Thanks,

Aharpster