Stable

LangGraph Integration

Map planner, tool, and synthesis steps to decorated functions so LangGraph execution can be visualized and estimated in Neurovn.

Install & Run

Install

Setup
pip install langgraph langchainpip install neurovn

Run

Execute
NEUROVN_API_URL=https://agentic-flow.onrender.com python langgraph_app.py

Architecture Flow

Planner

Planner node decides next action and routes to tool/agent branches.

Execution

Tool and model steps are traced as workflow edges with timing metadata.

Analysis

Neurovn computes graph type, critical path, and scenario estimates.

Implementation Snippets

Pattern

python
from neurovn import trace@trace.agent(name="Planner", model="gpt-4o")def planner(state: dict) -> dict:    return state@trace.tool(name="Retriever")def retrieve(query: str) -> str:    return "docs"

Troubleshooting

Wrap `app.invoke(...)` or `app.stream(...)` in `with trace.session(..., source="decorator", canvas_name="...")` so one LangGraph run becomes one named Neurovn canvas.

Track retry edges explicitly in your graph state to reflect loop behavior in estimates.

Use condition node routing metadata to keep branch-path estimates interpretable.

Set `expected_calls_per_run` for orchestrator-heavy agents where applicable.

Related Integrations

Backend contracts: `/api/estimate`, `/api/traces/sessions`, `/api/canvases`