Web App Playwright Testing

Problem

Local web app testing is rarely blocked by missing frameworks; it is blocked by engineering details: starting one or more services, cleaning up ports and processes, finding selectors after dynamic rendering, and deciding when the DOM is safe to inspect. Reading large helper scripts into context or hand-writing server lifecycle for every task slows the session and increases errors.

How It Works

The skill separates concerns: scripts/with_server.py is treated as a black box for server startup and cleanup, while automation scripts only contain synchronous Playwright logic. Check --help before use, then start a single server or multiple services. Prefer sync_playwright(), descriptive selectors (text=, role=, CSS, ID), and explicit waits such as page.wait_for_selector(). Selector discovery follows a reconnaissance-then-action loop: inspect the rendered DOM, identify the node or role to target, then perform the action. For dynamic pages, wait for page.wait_for_load_state('networkidle') before checking the DOM; examples/ includes patterns for discovering buttons, links, and inputs, local HTML automation, and console logging.

Notes

Do not read scripts/ as documentation unless --help is insufficient; close the browser after tests. For login state, WebSocket updates, slow third-party APIs, or pages where networkidle never stabilizes, use more specific element waits, timeouts, or custom probes.