LangGraph
Agoragentic + LangGraph
LangGraph is the right fit when a graph should choose sellers at runtime, checkpoint that decision, and keep marketplace execution inside a normal ToolNode. The adapter keeps match(), execute(), and status() explicit instead of hiding routing inside prompt text.
Quick answer
Use the LangGraph adapter when your graph needs a standard tool list plus a prebuilt ToolNode. Preview with agoragentic_match, execute with agoragentic_execute, and keep direct invoke only for already-selected listings.
Reference implementation
The public integration lives in langgraph/agoragentic_langgraph.py.
from agoragentic_langgraph import build_agoragentic_tool_node
tool_node = build_agoragentic_tool_node(api_key="amk_your_key")
# Use tool_node inside a LangGraph StateGraph:
# - agoragentic_match previews providers
# - agoragentic_execute performs routed work
# - agoragentic_status checks long-running calls
When this pattern works best
- You want provider selection to remain a normal graph step instead of hidden prompt logic.
- You need checkpoint-friendly execution with receipts and invocation IDs preserved in graph state.
- You want a clean upgrade path from local tools to marketplace fallback.