Windows TUI paste loses Turkish characters (CP857 vs UTF-8)

Summary
Codex CLI interactive TUI (chat) loses Turkish characters (ç, ğ, ı, İ, ö, ş, ü) when pasting from clipboard on Windows. Same text via stdin works fine.

Environment

• OS: Windows 11 Pro 64-bit, build 26220
• PowerShell: 7.6.0-preview.5
• Codex CLI: 0.56.0
• Console code page (initial): 857 (Turkish DOS)
• Tried: chcp 65001, [Console]::InputEncoding/OutputEncoding = UTF8 (no change in TUI)
• Terminal: Windows Terminal/ConHost

Steps to Reproduce

1. Run codex to open TUI.
2. Copy Turkish text like: çğıöşü ÇĞİÖŞÜ ... (or a real paragraph with diacritics).
3. Paste into TUI.

Expected

• Pasted text should preserve all Turkish characters.

Actual

• Some Turkish characters are missing/corrupted (especially ı/İ/ğ/ş/ç/ö/ü).

Control Test (stdin works)

• Get-Clipboard -Raw | codex exec --skip-git-repo-check --color never -
• Output preserves characters correctly.

Notes / Hypothesis

• TUI reads console input through ANSI code page (CP857) path and converts to UTF-8, dropping diacritics. Possibly falling back to ReadConsoleA vs ReadConsoleW.
• Suggest ensuring Unicode path for input on Windows (ReadConsoleW/ConPTY UTF-16) or forcing UTF-8 via SetConsoleCP/SetConsoleOutputCP(65001) at startup.

Workarounds

• Provide initial prompt as arg: $p=Get-Clipboard -Raw; codex -- "$p"
• Use stdin: Get-Clipboard -Raw | codex exec --skip-git-repo-check -
• Resume with stdin: Get-Clipboard -Raw | codex exec resume --last -

Thanks!