E2E smoke-тест поведения агента в WP Rocket

★ 6.9 · testing

Writing end-to-end tests is a Claude Code skill that runs a single behavioral E2E probe — a smoke test of one primary scenario — for the grooming agent in the WP Rocket project, using Playwright MCP and curl. Reach for it when you need to quickly validate behavioral assumptions before writing a spec: does the current feature actually behave as described in the issue, what does the AJAX endpoint return for the scenario being changed, does the settings page load cleanly after toggling an option. The skill boots a local WordPress environment via bash bin/dev-up.sh (idempotent), then exercises exactly one happy-path scenario: backend, REST, and AJAX calls are verified with curl; UI interactions (settings page, dashboard notices) go through Playwright MCP with navigation and snapshot inspection. Output is a JSON object with status PASS/FAIL/SKIP and a plain-language summary. The direction is behavioral verification of the existing system before spec authoring — not writing or committing Playwright spec files (that belongs to the qa-engineer + e2e-qa-tester tier). No spec files or screenshots are published. Designed exclusively for the grooming-agent at planning time; implementation agents (backend-agent, frontend-agent) do not invoke this skill. Scoped to the wp-media/wp-rocket project running at localhost:8888.