langchain

Neo — LangChain Integration

Use Neo's MCP server as a toolset inside LangChain agents. Neo executes AI/ML workloads locally on the user's machine — files are written directly to their workspace, never to a remote server.

MCP server: https://mcpserver.heyneo.com/mcp Auth: Authorization: Bearer sk-v1-YOUR_KEY

---

Option A: MCP Toolset (recommended)

LangChain's langchain-mcp-adapters package loads all 8 Neo tools from the MCP server automatically.

import asyncio
import os
from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent
from langchain_openai import ChatOpenAI

async def main():
    async with MultiServerMCPClient(
        {
            "neo": {
                "url": "https://mcpserver.heyneo.com/mcp",
                "transport": "streamable_http",
                "headers": {"Authorization": f"Bearer {os.environ['NEO_SECRET_KEY']}"},
            }
        }
    ) as client:
        neo_tools = client.get_tools()

        agent = create_react_agent(
            ChatOpenAI(model="gpt-4o"),
            neo_tools,
            prompt=(
                "You are an AI assistant with access to Neo, a local AI/ML execution backend. "
                "Files are written directly to the user's machine — never to a remote server. "
                "Use neo_submit_task for any ML/AI work. "
                "Poll with neo_task_status until COMPLETED, then call neo_get_messages. "
                "Always pass workspace as the project root (git root), never a subdirectory."
            ),
        )

        result = await agent.ainvoke({
            "messages": [("human", "Train a churn prediction model on churn.csv")]
        })
        print(result["messages"][-1].content)

if __name__ == "__main__":
    asyncio.run(main())

Install:

pip install langchain-mcp-adapters langgraph langchain-openai
export NEO_SECRET_KEY=sk-v1-...
export OPENAI_API_KEY=...

---

Option B: Custom tools (no MCP dependency)