Your way: the status line, output styles and built-in teachers

The last lesson of the level is about making Claude Code work the way that suits you.

The status line

At the bottom of Claude Code there is a bar — the status line. What it shows is your call: Claude Code runs your script, feeds it JSON about the session, and displays whatever the script prints.

The quickest route is to ask for the script:

/statusline show the model, the folder name and what percentage of the context is used

Or build it by hand. The script reads the JSON and prints one line:

#!/bin/bash
input=$(cat)
MODEL=$(echo "$input" | jq -r '.model.display_name')
DIR=$(echo "$input" | jq -r '.workspace.current_dir')
PCT=$(echo "$input" | jq -r '.context_window.used_percentage // 0' | cut -d. -f1)
echo "[$MODEL] ${DIR##*/} | ${PCT}% context"

And it is hooked up in your personal settings, ~/.claude/settings.json:

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/statusline.sh"
  }
}

The context percentage in the status line is /context from the first level, always in view.

Output styles

A style changes the way Claude answers:

/output-style learning

• Explanatory — Claude adds short Insight blocks explaining why it did things the way it did.
• Learning — the same, plus it leaves you a few lines of key decisions to write: it marks a TODO(human) spot and stops, asking you to fill it in.

The choice is saved in the project’s .claude/settings.local.json. To go back to the normal manner, use /output-style default.

Built-in teachers

• /powerup — short interactive lessons on Claude Code’s features, right in the terminal.
• claude-code-guide — the built-in helper for Claude Code itself. Name it in your request: “use the claude-code-guide agent: how do I make my own slash command?” Otherwise Claude may answer from memory, and memory goes stale faster than the docs.