Bug: Copying Japanese text from chat output results in mojibake (UTF-8 misinterpreted as Latin1)

[flanny7]

Description

Description

When copying Japanese text directly from the chat output in OpenCode, the text becomes mojibake. The same text displays correctly on screen, but becomes corrupted only when copied to the clipboard.

This issue appears to be specific to the copy operation within the OpenCode UI.

Steps to Reproduce

1. Open OpenCode and interact with the AI in Japanese.
2. Ask the AI to output Japanese text (e.g. a commit message, PR title, or code comment).
3. Select the Japanese text from the chat output and copy it (using mouse selection or any built-in copy functionality).
4. Paste the text into any application (terminal, editor, etc.).

Expected Behavior

The Japanese text should be copied correctly as UTF-8.

Example of correct text:

feat(calendar): 日詳細ボトムシート（P3）を追加

Actual Behavior

The copied text is garbled. It appears that the UTF-8 byte sequence is being misinterpreted as ISO-8859-1 / Windows-1252.

Example of actual copied result:

feat(calendar): æ¥è©³ç´°ããã ã·ã¼ãï¼P3ï¼ã追å 

Environment

• OpenCode version: 1.15.12
• OS: macOS
• Terminal / Environment: VS Code integrated terminal
• Locale: Japanese (Japan)
• How you are copying: Mouse drag selection

Additional Context

• The text renders correctly in the OpenCode chat UI.
• Copying the same Japanese text from other sources (files, browser, other applications) works normally.
• This only occurs when copying text that was output by the AI inside OpenCode.
• Related issues that may be connected:
  • fix: CJK characters garbled in TUI on Windows with non-UTF8 locale #14768 (CJK characters garbled in TUI on Windows)
  • Fix CJK character display issues in Windows terminals #10491 (Fix CJK character display issues in Windows terminals)
  • fix: PowerShell output encoding for non-ASCII characters on Windows #23636 (PowerShell output encoding issues on East Asian locales)

This seems to be a problem in the clipboard / text extraction layer rather than the rendering layer.

Plugins

No response

OpenCode version

1.15.12

Steps to reproduce

1. Open OpenCode and interact with the AI in Japanese.
2. Ask the AI to output Japanese text (e.g. a commit message, PR title, or code comment).
3. Select the Japanese text from the chat output and copy it (using mouse selection or any built-in copy functionality).
4. Paste the text into any application (terminal, editor, etc.).

Screenshot and/or share link

No response

Operating System

macOS 26.5

Terminal

VS Code integrated terminal

[garam-park]

#31126 (comment)

[garam-park]

ref microsoft/vscode#319821

[garam-park]

This appears to be a VS Code Terminal / OSC52 issue rather than an OpenCode issue.

I reproduced the same behavior using a raw OSC52 sequence outside OpenCode.

I also verified that clipboard copy works correctly in Antigravity.

Environment:

• VS Code 1.123.0
• SSH session
• Korean and Japanese text become mojibake
• ASCII text copies correctly

[igugyj]

[Bug]：VSCode-opencode 复制中文乱码 Copying Chinese text from VSCode-opencode results in garbled characters #31311

[loyalty-fox]

so is there any solution?

[igugyj]

so is there any solution?

open your clipboard then paste. #31311

it's not like a solution but a joke.

[dmsthd]

I am experiencing the exact same issue with Korean text on the latest version v1.17.3.
This is a duplicate of my previous issue #31978, which I have closed to centralize the discussion here.

Just like the Japanese text issue described here, English text copies perfectly, but Korean text becomes corrupted/garbled upon copying to the clipboard from the chat output. This indicates that the CJK/UTF-8 encoding regression is still present in the recent releases and affects multiple platforms.

My Environment:

• OpenCode Version: 1.17.3
• OS: Windows 11 (WSL - Windows Subsystem for Linux)
• Terminal: VS Code Integrated Terminal (bash)
• How you are copying: Mouse drag selection / Clipboard copy