Error: stream disconnected before completion: stream closed before response.completed

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #9936
• #9635
• #9727
• #9830

Powered by Codex Action

Thanks for uploading the logs. I don't see any smoking guns in the logs.

This looks to me like a vanilla networking connectivity issue. Are you using Codex on a corporate network? If so, I recommend working with your IT team to investigate issues with routing, proxies, firewalls, etc.

This worked for me:

The issue was a stale, cached authentication session in the local config (it was prioritizing an old/limited API Org instead of my ChatGPT subscription).

1. Kill lingering processes and remove the cached auth file:

pkill -f codex
rm ~/.codex/auth.json

1. Start fresh (forcing no env vars to ensure a clean login):

env -u OPENAI_API_KEY -u OPENAI_ORGANIZATION codex

1. Select "Sign in with ChatGPT" when prompted.

Lmk if it also works for you or helps?

here is a new one.
Or mention your thread ID 019c13b5-b3e2-74b2-aa13-f351fb90a614 in an existing issue. thank you! great stuff codex, I like it more than claude code :)). for now.

@ibabencu, thanks for uploading your logs. That error occurred because you tripped a server-side check that blocks conversations that are considered unsafe. I don't see anything in your prompt that would be an obvious cyber risk or similar, so this was likely a false positive. We're always working to better tune these safety detectors. If you see other instances of this, please use /feedback to upload the session details. We make use of this data to improve the models and detectors.

I've uploaded a /feedback, thread ID 019c2dad-c1c0-7331-bb4b-0e19534b637a.

[like] Ionut Babencu reacted to your message:
________________________________
From: Anton Timmermans @.*>
Sent: Thursday, February 5, 2026 1:53:40 PM
To: openai/codex @.*>
Cc: Ionut Babencu @.>; Mention @.>
Subject: Re: [openai/codex] Error: stream disconnected before completion: stream closed before response.completed (Issue #9995)

[https://avatars.githubusercontent.com/u/584693?s=20&v=4]atimmer left a comment (openai/codex#9995)<https://github.com/openai/codex/issues/9995#issuecomment-3853814382>

I've uploaded a /feedback, thread ID 019c2dad-c1c0-7331-bb4b-0e19534b637a.

—
Reply to this email directly, view it on GitHub<https://github.com/openai/codex/issues/9995#issuecomment-3853814382>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AWJOXJ35N7GCTCYN3BZBUQL4KNDOJAVCNFSM6AAAAACTBH6276VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTQNJTHAYTIMZYGI>.
You are receiving this because you were mentioned.Message ID: @.***>

@atimmer, thanks for uploading your logs. I looked at them, and I suspect the issue you're hitting is due to very long tool outputs. We'll need to look into it further. In the meantime, as a potential workaround, try to reframe the task or adjust your skills / MCPs so the output of tool calls is smaller.

Using the VSCode plugin, I logged out by clicking the settings cog icon at the top of the view and logged back in

I kept running into this on Azure with gpt-5.3-codex and 5.4. Built a workaround that auto-retries on stream disconnect and splits large tasks into separate sessions to avoid context overflow and compaction errors.
https://github.com/pauldinu10/codex-smart
It's a PowerShell wrapper that runs each task from a tasks.md file as a fresh codex exec --full-auto session. If the stream drops, it retries automatically up to N times. Not a fix, but makes Codex usable for long migrations.

stream disconnected before completion: stream closed before response.completed