Playwright

Agents that interact with the web need programmatic control over navigation, clicks, form fills, and screenshots. Playwright is faster and more reliable than Selenium, with auto-waiting, network interception, and headless/headed modes.

Playwright is a remote control for a web browser — your agent presses buttons, reads screens, and navigates pages as if a human were sitting at the computer.

Scenario

Agent navigates to a SaaS dashboard, logs in with stored credentials, clicks 'Export CSV', waits for download, and parses the file for analysis.

Solution

In Browser & Computer Use Agents, apply Playwright to this scenario: Agent navigates to a SaaS dashboard, logs in with stored credentials, clicks 'Export CSV', waits for download, and parses the file for analysis. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to Playwright (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where Playwright happens in the code.

1from playwright.async_api import async_playwright  # import dependencies2
3async def scrape_pricing(url: str) -> str:4    async with async_playwright() as p:  # key line for Playwright5        browser = await p.chromium.launch(headless=True)6        page = await browser.new_page()7        await page.goto(url)8        await page.wait_for_selector("[data-testid='price']")9        price = await page.locator("[data-testid='price']").text_content()10        await browser.close()11        return price  # return the result

• Treating Playwright as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for playwright

Quick recap — the most important points from this module.