Approve Claude Code commands without switching windows

Claude Code stops and waits when it needs permission. Here's how to get alerted the second it asks, and how to Allow or Deny without going back to the terminal.

Updated October 15, 2026

Before Claude Code runs a shell command, edits a file outside the rules you set, or fetches from the web, it can stop and ask. That's a good safety net, but it has a cost: the whole task stops until you answer. If you're in another window, Claude may wait a long time for a one-click “yes”.

1. Get alerted the moment Claude asks

Claude Code fires a Notification hook when it needs your permission. With a matcher of permission_prompt, you can alert only on permission requests and stay quiet otherwise. Add this to ~/.claude/settings.json on macOS:

{
  "hooks": {
    "Notification": [
      {
        "matcher": "permission_prompt",
        "hooks": [
          {
            "type": "command",
            "command": "osascript -e 'display notification \"Claude Code is waiting for your permission\" with title \"Claude Code\" sound name \"Ping\"'"
          }
        ]
      }
    ]
  }
}

Start a new session after editing the file. You'll get a notification, but you still need to switch back to the terminal to answer.

2. Cut down how often it asks

If you approve the same safe commands again and again, such as your test runner or linter, add them to the allow list in your Claude Code permission settings (run /permissions in Claude Code). Keep anything destructive on “ask”. Fewer prompts means fewer stalls, and the ones that remain are the ones worth reading.

3. Answer right from your screen edge

Flincko shows Claude's permission request in a bubble on its desktop buddy, with the command it wants to run. Click Allow or Deny right there, or jump to the terminal if you want to look closer. No window switching, and the task keeps moving.

More guides