[BUG] CLI crashes with panic on UTF-8 emoji in CLAUDE.md

Bug Description

The Claude Code CLI crashes with a Rust panic when processing emoji characters in the CLAUDE.md file. The error occurs during UTF-8 string processing when the code attempts to split a string at a non-character boundary inside a multi-byte emoji character.

Error Message

thread '<unnamed>' (19268) panicked at /rustc/ed61e7d7e242494fb7057f2657300d9e77bb4fcb/library/core/src/str/mod.rs:833:21:
byte index 3 is not a char boundary; it is inside '🔴' (bytes 0..4) of `🔴'}</span>`
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
fatal runtime error: failed to initiate panic, error 5, aborting
zsh: abort      claude

Steps to Reproduce

1. Create a CLAUDE.md file in your project root
2. Add emoji characters (e.g., 🔴, 🟡, 🟢) anywhere in the file
3. Run the claude command
4. The CLI immediately crashes with the panic error above

Expected Behavior

The CLI should handle UTF-8 multi-byte characters (including emoji) correctly without crashing.

Actual Behavior

The CLI crashes with a panic when attempting to process emoji characters, specifically when trying to index into the string at a byte position that falls within a multi-byte character boundary.

Environment

• Platform: macOS (Darwin 24.1.0)
• Claude Code Version: Latest (based on error timestamp)
• Shell: zsh

Workaround

Remove all emoji characters from CLAUDE.md and replace them with plain text alternatives.

Example fix:

- ### 🔴 Critical
+ ### [CRITICAL]

Root Cause Analysis

The error indicates that the CLI is attempting to slice a UTF-8 string at byte index 3, which falls inside the 4-byte emoji character '🔴'. This suggests the code is using byte indexing instead of character-aware string operations.

Suggested Fix

Use character boundary-aware string operations when parsing CLAUDE.md content. Rust's string slicing should use methods like:

• str::char_indices() for iterating with proper boundaries
• str::get(range) instead of direct indexing, which returns None for invalid boundaries
• Ensure all substring operations respect UTF-8 character boundaries

Additional Context

This is a critical bug as it prevents users from using emoji in their documentation files, which is increasingly common for visual clarity and categorization.