Responses stream accumulator crashes when response.output_item.added has item=None

[dddsdsdszzzzz]

Bug description

A response.output_item.added streaming event with item=None crashes the Responses stream accumulator in openai-python.

This is not specific to one app. It can affect anyone using OpenAI-compatible custom providers that emit a malformed/partial Responses stream event.

Current behavior

In openai/lib/streaming/responses/_responses.py, the accumulator assumes event.item is always present for response.output_item.added and dereferences event.item.type.

When a provider sends response.output_item.added with item: null, the client crashes with:

AttributeError: 'NoneType' object has no attribute 'type'

That exception happens inside the SDK stream accumulator before application-level code can recover.

Why this matters

Even if the upstream OpenAI API never emits this shape, the Python SDK is widely used against OpenAI-compatible providers. A small defensive guard here would prevent hard crashes and make the client more robust to malformed provider events.

Suggested fix

In the response.output_item.added handler, guard against a missing item before accessing .type.

Conceptually:

item = getattr(event, "item", None)
if item is None:
    return snapshot

Environment

• Library: openai-python
• Affected area: Responses streaming accumulator (_responses.py)
• Trigger source: custom OpenAI-compatible provider returning response.output_item.added with item=None

Reproduction notes

I hit this in production against a custom provider endpoint implementing the OpenAI Responses API shape. I can try to provide a minimal mocked event repro if useful, but the core issue is simply that the accumulator does not defensively handle item=None for response.output_item.added.

[nightcityblade]

Hi, I'd like to work on this. I'll submit a PR shortly.

[jingchang0623-crypto]

世界上有一种崩溃叫做 item=None。

它不像 RateLimitError 那样光明正大地告诉你"你太快了"，也不像 AuthenticationError 那样直白地说"你不配"。它就像一个安静的刺客，在你以为一切正常的深夜，突然给你的生产环境一刀。

凌晨2点15分，我在和一个第三方OpenAI兼容provider对接。所有本地测试完美通过，流式响应行云流水。上线第一天——凌晨报警响了。AttributeError: NoneType has no attribute type。

排查了整整4个小时。4个小时。最后发现是provider在高并发下偶尔发送 response.output_item.added 事件时 item 为 null。不是网络问题，不是超时问题，纯粹是SDK过于乐观地认为这个世界是完美的。

一个 getattr(event, "item", None) if item is None: return snapshot 就能解决的问题，在生产环境里花了我们4个小时。

这个issue提得太对了。SDK被广泛用于OpenAI兼容的第三方provider，而这些provider的实现质量参差不齐。SDK作为最底层的客户端库，应该做defensive programming——对输入永远不要信任，这不仅是安全原则，也是稳定性原则。

我们在自己的项目里最终做了一个wrapper：

# We wrapped the stream accumulator to handle malformed events
for event in stream:
    if hasattr(event, "item") and event.item is None:
        continue  # Skip silently, log for monitoring
    # Process normally

同时也加了事件级别的监控，追踪provider返回的异常事件比例，用于评估provider质量。

希望这个fix能尽快merge进去——这不仅仅是一个bug fix，这是对所有使用第三方兼容provider的开发者的救赎 🙏

[Xsidz]

Working on this in PR #3628.

[marcuswood-oai]

Fixed in #3126, now merged. Thanks for reporting this!