OPENAI_API_KEY is not set, skipping trace export

[lonngxiang]

from agents import Agent, Runner, AsyncOpenAI, OpenAIChatCompletionsModel, ModelSettings, handoff
from agents.extensions.handoff_prompt import prompt_with_handoff_instructions
import asyncio
import json
from pydantic import BaseModel, ValidationError

# 设置OpenAI客户端
openai_client = AsyncOpenAI(base_url="https://api.deepseek.com", api_key="sk-******")

[WSQsGithub]

You need to export env var explicly in your terminal using export OPEN_API_KEY=sk-xxxx or use load_env() to do that.
It seems like agents-sdk doesn't have an interface like openai sdk.

[Salaudev]

@lonngxiang

if you want to use an external LLM, you need to disable tracing

[EnggQasim]

from agents import set_tracing_disabled
set_tracing_disabled(disabled=True)

Just apply this code.

[jaijuneja]

Just wanted to add to this that I get the same error when using python_dotenv. I'm using OpenAI's models, not a 3P model.

I set OPENAI_API_KEY in my .env file and then have this at the top of my code:

from dotenv import load_dotenv

load_dotenv(override=True)

# Rest of code

This still raises the same warning message about skipping trace export.

I've found two ways to address this:

1 - set the environment variable explicitly in the terminal (export OPENAI_API_KEY=<key>)

2 -

import os
from dotenv import load_dotenv

load_dotenv(override=True)

OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")

if OPENAI_API_KEY:
    from agents import set_tracing_export_api_key
    set_tracing_export_api_key(OPENAI_API_KEY)

[rm-openai]

@jaijuneja good point. #289 should fix it.

[Leon-Sander]

@rm-openai is it already in the main branch? Still getting the error

[rm-openai]

Its merged but not released, the next version (v0.0.7) will have it

[sgb-io]

This fix is available now -- https://github.com/openai/openai-agents-python/releases/tag/v0.0.7