LangGraph Integration
Foxhound integrates with LangGraph and LangChain via a callback handler that captures every step — LLM calls, tool invocations, and chain transitions.
Installation
pip install foxhound-ai[langgraph]
Usage
from foxhound import FoxhoundClient
from foxhound.integrations.langgraph import FoxCallbackHandler
fox = FoxhoundClient(api_key="fox_...", endpoint="https://your-foxhound-instance.com")
handler = FoxCallbackHandler.from_client(fox, agent_id="my-langgraph-agent")
result = await graph.ainvoke(state, config={"callbacks": [handler]})
await handler.flush()
Pass the FoxCallbackHandler in the callbacks list of the LangGraph/LangChain invocation config. The handler intercepts every framework lifecycle event and sends it to Foxhound as structured spans.
What gets traced
- LLM calls (model name, prompt, completion, token counts)
- Tool calls (tool name, inputs, outputs)
- Chain start/end with intermediate steps
- Agent actions and observations
- Errors with full stack traces
Why use the native integration?
The FoxCallbackHandler has direct access to raw LangGraph/LangChain event payloads, capturing per-turn tool call arguments and intermediate reasoning steps that the OTel bridge cannot access from generic GenAI spans.