[Bug] Garbled CJK characters in OpenCode console due to Windows system default ACP (GBK/CP936) encoding #31187

Description

Environment

• OS: Windows 11 (Simplified Chinese edition)
• Shell: PowerShell 7
• System active code page: 936 (GBK)

Problem Description

The OpenCode console inherits the system's default Active Code Page (ACP) encoding in Windows. On Simplified Chinese versions of Windows, the default system ACP is 936 (GBK). New child processes created by OpenCode inherit this encoding setting for standard input and output streams.

When a command-line tool (or child process) outputs text encoded in UTF-8, the parent OpenCode console reads these UTF-8 bytes using the system's default GBK decoder, resulting in garbled characters (mojibake). This issue primarily affects CJK (Chinese/Japanese/Korean) characters, while pure ASCII characters display normally.
the console output: CJK characters are rendered as garbled symbols, while English/ASCII text is displayed correctly.

Root Cause

• The child process's standard output is encoded in UTF-8.
• The parent OpenCode console uses the system's default GBK (CP936) decoder to interpret the UTF-8 byte stream, causing misinterpretation of multi-byte CJK characters.

Plugins

none

OpenCode version

1.15.13

Steps to reproduce

1. Run OpenCode in a Simplified Chinese Windows 11 environment with PowerShell 7.
2. Execute a script/command that outputs UTF-8 encoded text containing CJK characters (e.g., the PowerShell script in the screenshot reading UTF-8 text files).
3. Observe the console output: CJK characters are rendered as garbled symbols, while English/ASCII text is displayed correctly.

Screenshot and/or share link

Operating System

Windows 11

Terminal

Windows Terminal, powershell 7