Playwright V2

Playwright is the current gold standard for browser automation and E2E testing — superior auto-waiting (no flaky sleep() calls), multi-browser support (Chromium, Firefox, WebKit), and an excellent Python API. For agents: automate web interactions (fill forms, click buttons, navigate pages), extract web content (scrape dynamic pages), generate PDFs and screenshots, and run E2E test suites. Playwright Test provides parallel execution, tracing, and HTML reports. No server component — runs fully locally. Confidence is docs-derived.

Python API (playwright-python). sync API: with sync_playwright() as p: browser = p.chromium.launch(headless=True); page = browser.new_page(); page.goto("https://example.com"); page.fill("#username", "user"); page.click("#submit"); result = page.text_content(".result"); screenshot = page.screenshot(path="screenshot.png"); browser.close(). Async API (async with async_playwright()). Page methods: page.goto(url), page.fill(selector, value), page.click(selector), page.select_option(selector, value), page.wait_for_selector(selector), page.evaluate(js_expression), page.screenshot(), page.pdf(). Network: page.route(pattern, handler) intercepts requests. Playwright Test (JS/TS): test("name", async ({page}) => { ... }) with built-in fixtures. Code generation: playwright codegen {url} records user actions as test code.

No external auth — Playwright controls browsers directly. Browser download: playwright install downloads browser binaries (Chromium ~300MB). Docker: mcr.microsoft.com/playwright image for CI with all browsers pre-installed. Headed vs. headless: headless=True for CI; headed for debugging. Browser context: browser.new_context() for isolated sessions (separate cookies/storage). HTTPS: Playwright handles TLS automatically — set ignore_https_errors=True for self-signed certs in testing.

Auto-waiting: Playwright waits for elements to be actionable before interacting — eliminates most race conditions. Selector strategies: CSS, text, accessibility role, XPath. Strict mode: single-element selectors fail if multiple elements match — prevents silent errors. Flaky test detection: Playwright Test --retries for automatic retry on failure. Trace viewer: trace.zip recorded per test — open in Playwright trace viewer for debugging. Browser processes: each new browser instance is a separate process — manage lifecycle carefully for resource-intensive agents. Screenshot on failure: built-in for Playwright Test.

playwright.dev covers Python/JS/TS API reference, browser automation guide, Playwright Test documentation, CI/CD integration, and tracing guide. Getting started: pip install playwright && playwright install — first browser automation in under 5 minutes. Official Docker image for CI. Community via Playwright Discord (active) and GitHub (70k+ stars). Exceptional documentation quality with interactive demos. One of the most rapidly growing testing tools.