When a Playwright Script Should Become a Browser Skill

A Playwright script usually starts as a small fix. You need to check a page. You need to open a dashboard. You need to extract a value. You need to confirm whether a logged-in account still works. So you write a script. It launches a browser, visits a URL, waits for a selector, clicks a button, prints a result, and exits. For a local test, that is often enough. The trouble starts later. The same script gets copied into another folder. Someone adds a proxy flag. Someone else changes the account. A teammate wraps it inside an AI agent workflow. A scheduled job starts running it at night.