Tunnel information incorrect for Cloudflare tunnel

Resolved 💬 3 comments Opened Jan 30, 2026 by avirweb Closed Jan 30, 2026

💡 Likely answer: A maintainer (etraut-openai, contributor) responded on this thread — see the highlighted reply below.

What version of Codex is running?

0.92

What subscription do you have?

Pro

Which model were you using?

Codex 5.2

What platform is your computer?

_No response_

What terminal emulator and version are you using (if applicable)?

_No response_

What issue are you seeing?

Proxy has to be off. Had me chasing DNS for an hour which was wrong. Proxy off for tunnels!! Kept saying ON!

Also, why does every AI use 8000 on the same machine without checking!

What steps can reproduce the bug?

Uploaded thread: 019c0f61-0c84-7ff1-a2c0-5c9f11129513

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

3 Comments

etraut-openai contributor · 6 months ago

Can you provide more context? Which proxy are you referring to? Can you describe your network configuration? Do you have repro steps?

avirweb · 6 months ago

Pretty much blank domain. I don't want to put the domain name here but anyway it's a cloudflare tunnel https://developers.cloudflare.com/cloudflare-one/networks/connectors/cloudflare-tunnel/ and it requires a DNS record to be setup. For whatever reason, the AI thinks that on dig and getting a cloudflare proxy IP is incorrect when it is correct. It keeps saying DNS with a proxy ip will return a cname record like <tunnel>.domain.com

Honestly, DNS and networking and ports and private networking and IPv4, IPv6 really needs to be more trained up on ALL AI not just Codex. Even Opus 4.5 gets really stupid about port numbering and private networks like zerotier.

Codex was saying earlier no Zerotier configured even as I could ping the ZT IP address.

IDK how else to say but these AI are just not network engineers.

etraut-openai contributor · 6 months ago

Thanks for the bug report. We'll share the feedback with the team responsible for training our codex models. We're tracking this feedback and similar reports so we can make future models more robust and efficient. And thanks for using the /feedback command. That additional information is useful.