Feature request: visually distinguish user prompts from assistant output in CLI transcript

Problem

When scrolling back through a long Claude Code CLI session, it is very hard to locate where my own prompts (questions) were. The visual hierarchy is too flat:

• User prompt
• Tool calls / status lines
• Assistant response
• Next user prompt
• ...

All of these blend together in the terminal scrollback. To re-read a question I asked earlier, I currently have to scan line-by-line looking for it. There is no strong visual anchor that says "this is where the user spoke."

Suggestion

Please make the user prompt block visually distinct from tool activity and assistant output. Some ideas:

1. A colored left-margin bar or gutter (e.g. a bright vertical bar) next to every user message, so scrolling reveals prompt locations at a glance.
2. A prompt header line like ── you ───────────── above each user message, similar to how many chat CLIs (e.g. aider, chatgpt-cli) separate turns.
3. Different background/foreground color for user text vs. assistant text vs. tool output.
4. Optional numeric turn index ([#12] you: / [#12] claude:) so I can reference "go back to turn 12".
5. A slash-command like /prompts that lists all user prompts in this session with their turn number, so I can jump back.

Any one of these would already be a huge improvement. Ideally the CLI would ship 1+2 by default.

Why this matters

Long sessions (hours, many tool calls) are the norm. Being able to re-audit "what did I actually ask?" is essential — right now the question I asked 30 minutes ago is buried under hundreds of lines of tool output and indistinguishable from everything else.

Thanks for considering.