What problem does it solve? Creating user guides and manuals requires real screenshots of a running application, but capturing them is error-prone: local multi-tenant or reverse-proxied apps reject requests with the wrong Host header, login flows need automation, and fragile selectors silently capture the wrong page. This Skill provides a proven Playwright-based workflow that handles these pitfalls. ## Core Features & Use Cases - Headless browser capture: Drives Chromium via Playwright to log in, navigate, and screenshot real application screens on Windows, macOS, and Linux. - Multi-tenant/reverse-proxy handling: Uses Chromium's --host-resolver-rules flag to map a real domain to a local instance, solving Host-header routing problems common in self-hosted apps like Frappe/ERPNext. - Session reuse and credential hygiene: Saves login state via storageState per user role and keeps secrets out of command output and transcripts. - Selector robustness and visual debugging: Avoids silently-wrong captures through scoped selectors, explicit waits, and read-back verification of every screenshot. - Use Case: Document an ERPNext login and order-entry flow for a user manual by capturing each screen as a specific role, with ordered filenames ready to drop into a document. ## Quick Start Use the screen-capture skill to log into my local app and take screenshots of the login, dashboard, and order creation screens for a user guide.
