enable-voice-alerts

Enable Voice Alerts

This command configures macOS text-to-speech notifications for Claude Code events.

What It Does

Sets up voice alerts for:

• Permission requests - "Claude needs your permission"
• Questions - "Claude has a question for you"
• Idle prompts - "Claude is waiting for your input"
• Session completion - "Claude has finished"

Setup Process

1. Determine the Claude config directory:

   • If ~/.claude-work/ exists → use ~/.claude-work/settings.json
   • Otherwise → use ~/.claude/settings.json

2. Read the existing settings.json file

3. Add or update the hooks configuration:

{
  "hooks": {
    "PermissionRequest": [
      {
        "matcher": "*",
        "hooks": [
          {
            "type": "command",
            "command": "say 'Claude needs your permission'"
          }
        ]
      }
    ],
    "PreToolUse": [
      {
        "matcher": "AskUserQuestion",
        "hooks": [
          {
            "type": "command",
            "command": "say 'Claude has a question for you'"
          }
        ]
      }
    ],
    "Notification": [
      {
        "matcher": "idle_prompt",
        "hooks": [
          {
            "type": "command",
            "command": "say 'Claude is waiting for your input'"
          }
        ]
      }
    ],
    "Stop": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "say 'Claude has finished'"
          }
        ]
      }
    ]
  }
}

1. Preserve all existing settings - only add/update the hooks key

2. Write the updated settings.json

Completion Message

After setup, tell the user:

Voice alerts enabled!

You'll now hear spoken notifications when Claude:
- Needs permission for an action
- Has a question for you
- Is waiting for your input
- Finishes a task

Settings saved to: <settings_path>

To disable, remove the "hooks" section from your settings.json

Notes

• Platform: macOS only (uses say command)
• No restart needed: Hooks take effect immediately
• Customization: Users can edit settings.json to change the spoken phrases