[FEATURE] Add thousand separators to token count display

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

The token count displayed in the Claude Code prompt shows raw numbers without thousand separators (e.g., 45678 instead of 45,678). As context usage grows into the tens and hundreds of thousands of tokens, these numbers become hard to read at a glance.

Proposed Solution

Add locale-aware thousand separators to the token count display in the CLI.

For example:

• 45678 → 45,678
• 123456 → 123,456
• 1048576 → 1,048,576

Alternative Solutions

I tried adding a comma-formatted token count via a custom statusLine command in settings.json using printf "%'d", but this creates duplicate information since the native token count is still displayed without commas alongside the status line.

Priority

Low - Nice to have

Feature Category

Interactive mode (TUI)

Use Case Example

1. When monitoring context window usage during a long session, I glance at the token count to decide whether to start a new conversation. With large numbers like 458392, it takes a moment to parse — 458,392 is instantly readable.

1. The 1M context models (e.g., opus[1m]) have higher per-token costs. Token counts routinely reach six or seven digits in these sessions, making unformatted numbers like 734218 even harder to quickly parse. Being able to instantly read 734,218 helps with cost-awareness when every token costs more.

Additional Context

This issue was written by Claude Code on behalf of the author.