[BUG] Non-ASCII characters (Cyrillic, Chinese, Hindi, etc.) displayed as "?" or diamond symbols in Roo Code output

[jwadow]

Problem (one or two sentences)

When running commands that produce non-English characters (e.g., Cyrillic, Chinese, Hindi), the Roo Code agent in VS Code displays question marks (?) or diamond symbols with question marks instead of the correct characters.
The model seems to understand the text internally — only the output rendering is broken.

Context (who is affected and when)

Affects all users who use non-ASCII languages (e.g., Russian, Chinese, Hindi, Arabic, etc.) in Roo Code’s output panel.
The issue is visible when executing any command or script that outputs such text.
It does not occur in the regular terminal (CMD or PowerShell) — only in the Roo Code output terminal.

Reproduction steps

0. Windows 11, VS Code 1.104.3
1. Open VS Code with Roo Code extension enabled.
2. Ask Roo: please run command "echo hi, привет, 你好, नमस्ते"
3. Observe the Roo Code agent output — non-English text becomes ? or �.
4. Compare the same command in Windows CMD or PowerShell — all characters display correctly.

Expected result

All Unicode characters (e.g., привет, 你好, नमस्ते) should display correctly, matching terminal output.

Actual result

Roo Code output panel replaces all non-ASCII characters with ? or diamond symbols with question marks, making it impossible to read multilingual output.

Variations tried (optional)

Tried both CMD and PowerShell -> both show correct characters.
Roo Code output -> broken.
Issue persists regardless of system locale or font settings.
Confirmed on Windows 10/11, VS Code latest version.

App Version

3.28.15

API Provider (optional)

None

Model Used (optional)

No response

Roo Code Task Links (optional)

No response

Relevant logs or errors (optional)

[roomote]

I'm exploring a potential fix for the Unicode character display issue in the Roo Code output terminal. I'll investigate the encoding handling in the terminal output and open a PR to share progress.

[roomote]

Opened PR #8531 with my attempt at addressing this issue. The fix adds UTF-8 encoding support by setting locale environment variables and prepending the 'chcp 65001' command on Windows terminals. Feedback welcome!

PR: #8531