Codex use UTF-8 BOM by default and change Chinese to unreadable nonsenese

[UIZorrot]

What version of the Codex App are you using (From “About Codex” dialog)?

latest

What subscription do you have?

plus

What platform is your computer?

windows, wsl

What issue are you seeing?

just let codex or gpt know, please use UTF-8 with no BOM by default.

What steps can reproduce the bug?

let codex do some chinese work and you will see. May other utf-8 language have this issue.

What is the expected behavior?

just let codex or gpt know, please use UTF-8 with no BOM by default. And when using command like get-content, should use utf-8 for non-english language.

Additional information

like this

[github-actions]

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

• Codex patch french format files in UTF-1 instead of UTF-8 causing massive mojibake #14227
• Windows PowerShell UTF-8 Markdown files can render as mojibake in agent context, causing patch mismatch and unreliable edits #15422

Powered by Codex Action

[mbetrifork]

I prepared a fix for this on my fork, but this repository is currently configured to block PRs from non-collaborators

Branch:
https://github.com/mbetrifork/codex-fixes/tree/issue-15967

Commit:
a8eaccf Fix #15967: default PowerShell commands to UTF-8

Summary:

• defaults PowerShell cmdlet encoding to UTF-8
• fixes the Get-Content/UTF-8-no-BOM read path that can garble Chinese and other non-ASCII text on Windows PowerShell
• adds regression tests

I can make a patch file if that's the only way to contribute the fix.

[lightcomc]

bad that utf-8 md files is mojibaked EVERY time in codex Powershell calls. Fix wanted