Hebrew in command support (ascii jumps)

[nave121]

What version of Codex is running?

codex-cli 0.65.0

What subscription do you have?

pro

Which model were you using?

5.1

What platform is your computer?

Darwin 25.0.0 arm64 arm

What issue are you seeing?

When trying to work with a Hebrew file name, it glitches to gibberish

What steps can reproduce the bug?

Uploaded thread: 019aedc3-6898-75f2-9aac-f51f5ee010c4

What is the expected behavior?

Ran cd 'examples/not orgenaized yet/fw' && touch 'IL-166-12-2024, 26106, ÎûÎÖÎáÎÆο.html' && ls
└ /opt/homebrew/Library/Homebrew/cmd/shellenv.sh: line 18: /bin/ps: Operation not permitted
IL-166-12-2024, 26106, ÎûÎÖÎáÎÆο.html

it tries to get to a non-existent file, instead of IL-166-12-2024, 26106, ÎûÎÖÎáÎÆο.html it supposed to be IL-166-12-2024, 26106, זינגר.html

Additional information

ascii jumps in commands

[etraut-openai]

Thanks for the report. This looks like a Unicode encoding issue. I'm trying to determine whether the encoding problem occurred — whether it arrived with the incorrect encoding from the model or within the codex harness.

I tried to repro using the latest version of the CLI (0.71.0) using default settings, and I wasn't able to repro.

Does this problem happen consistently for you, or is it intermittent? If it's consistent, are there any specific repro steps you can provide? I tried creating a file named "זינגר.html" and ask the model to read it. It read the file and reported the file name back with correct Hebrew characters.

Do you remember which model you were using at the time? According to the uploaded session, you were using the openai model provider, but we don't record the model in the session metadata. (You can switch models in the middle of a session.)