OpenAI Agents SDK

Attach the Gnesi MCP server with MCPServerStreamableHttp and a bearer credential.

The OpenAI Agents SDK speaks MCP over streamable HTTP. There is no Gnesi toolkit package.

Bash

pip install openai-agents
export GNESI_CREDENTIAL="gnesi_..."
export OPENAI_API_KEY="..."

Python

import asyncio
import os

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async def main() -> None:
    async with MCPServerStreamableHttp(
        name="Gnesi",
        params={
            "url": "https://api.gnesi.io/api/mcp",
            "headers": {"Authorization": f"Bearer {os.environ['GNESI_CREDENTIAL']}"},
        },
    ) as server:
        agent = Agent(
            name="Gnesi",
            instructions="Read and send this person's Gnesi conversations. Wait when a message is in Needs you.",
            mcp_servers=[server],
        )
        result = await Runner.run(agent, "List my Gnesi conversations that need me.")
        print(result.final_output)

asyncio.run(main())

Authorization works with both major versions of the MCP Python package. Put a per-user token on a new MCPServerStreamableHttp for that request. The header value is the agent credential from Settings, Agent access.

The OpenAI API can also call the same URL as a remote MCP tool. Put the bearer token in that tool's authorization field: Bearer plus the credential. Keep the credential out of the agent definition you store.

Tool names are on MCP.