Automate Browser Tasks with Playwright Go

Playwright Go Automation Expert is a skill for browser automation using github.com/playwright-community/playwright-go. It guides scripting for scraping, testing, and interaction with dynamic websites, including SPA, React, and Vue interfaces that require a real browser. Its implementation rules focus on browser architecture, observable actions, explicit error handling, and controlled session behavior.

The central architecture launches one browser and creates a separate BrowserContext for each distinct task or session. Cookies, cache, and storage are isolated by context, avoiding the overhead of a new browser process for every task. The workflow closes pages, contexts, and browsers with defer and wraps critical routines in panic recovery that logs stack traces. Explicit action timeouts replace reliance on default timeout settings.

Logging uses go.uber.org/zap. Development logging uses a console-friendly development logger, while production logging uses structured JSON. Navigation, clicks, and inputs are logged with contextual fields such as the selector being used. The rules discourage fmt.Println for application logging and pair critical actions with a safe runner and Zap records.

Interaction guidance includes variable typing delays, curved mouse movement with jitter, viewport variation, per-context User-Agent rotation, and incidental scrolling or hovering during waits. These techniques aim to emulate human interaction, but strict anti-bot settings can still detect automation. The skill has no built-in CAPTCHA solving.

Use it when writing Go scripts to scrape, automate, or test a website, when debugging an existing Playwright implementation, or when dynamic content requires browser execution. It also addresses requests involving detection avoidance and human-like behavior.

Full browser instances consume substantial RAM and CPU, so the single-browser, multiple-context design is part of resource management. Default scraping and testing behavior is read-only; scripts designed to submit forms or modify data require that behavior to be made explicit. Context state does not persist to the host unless it is saved deliberately. Detection resistance should not be treated as a guarantee.

Inputs include a target site, the intended scraping or testing task, selectors, session identity requirements, and debugging needs. Outputs are Go browser automation scripts with isolated sessions, structured logging, timeout choices, and cleanup paths. The implementation playbook covers initialization, typing and clicking helpers, and cookie save/load scaffolding. Cookie serialization and file handling remain implementation work rather than a completed storage feature.

For visible debugging, the checklist calls for headless mode to be disabled and SlowMo enabled. A new identity receives a new context, proxy configuration, and rotated User-Agent. Critical actions receive SafeAction handling and logging.

The skill uses Playwright Go and Zap, with Chromium initialization in its example. Drivers and browsers are required; the source installation command is:

go run github.com/playwright-community/playwright-go/cmd/playwright@latest install --with-deps

API knowledge is used first, with Playwright Go documentation consulted for unknown errors, complex interception or authentication flows, or significant API changes. The attached implementation playbook provides detailed patterns for initialization, interaction, and session management.