Unicode ambiguous-width characters (◆, ← etc.) overlap adjacent text in CLI output

Description

Characters classified as Unicode "Ambiguous Width" (e.g. ◆, ←, ■) overlap with adjacent characters in Claude Code CLI output. The rendered text appears garbled because the terminal treats these as double-width (2 cells), but Claude Code's layout engine appears to calculate them as single-width (1 cell).

This is the same class of issue seen in Powerlevel10k's setup wizard, where it asks whether ◆ occupies one or two columns.

Steps to reproduce

1. Have Claude Code output text containing ambiguous-width Unicode characters, e.g. ◆任務完了◆ or ### ◆SNet道場◆
2. Observe that ◆ overlaps into the next character

Expected behavior

Characters should not overlap. The width calculation should match the terminal's rendering.

Environment

• OS: WSL2 (Ubuntu) on Windows
• Terminal: Windows Terminal (WT_SESSION is set)
• TERM: xterm-256color
• Locale: C.UTF-8
• Claude Code: CLI (latest)

Notes

• This likely affects all terminals that render ambiguous-width characters as fullwidth (2 cells), which is the default for CJK locales and many terminal emulators.
• A common solution is to use a library like East Asian Width aware string-width calculation (e.g. Node.js string-width with {ambiguousIsNarrow: false} option or similar).