From rldyour-browser
Валидация UI и пользовательских сценариев в браузере через Playwright MCP. Используй для: проверь UI, проверь в браузере, валидируй страницу, скриншот, регрессия, адаптив, бизнес-логика, проверь визуально. EN triggers: validate UI, browser check, e2e test, regression test, responsive check, business logic in browser, take screenshot, pixel-perfect validation, visual QA.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rldyour-browser:browser-validationThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Validate browser-facing work with real browser evidence, not assumptions. The goal is to prove the implementation is visually correct, functionally correct, and consistent with business logic.
Validate browser-facing work with real browser evidence, not assumptions. The goal is to prove the implementation is visually correct, functionally correct, and consistent with business logic.
User-facing reports stay in Russian unless requested otherwise. Store browser artifacts under browser/ and do not commit them.
Use this skill without waiting for explicit invocation when the task asks to:
browser/.If the validation exposes console, network, runtime, layout, hydration, or performance problems, hand off to browser-debug.
Use this skill after:
If the user asks for pixel-perfect, production-quality, or "проверь в браузере", this skill applies.
Validate the implementation across these layers when relevant:
Use mcp__plugin_rldyour-mcps_playwright__* tools provided by Playwright MCP:
browser/ for key states: initial, changed state, error/empty/loading state, desktop/mobile, and final state.browser-debug).browser/ unless the user asks to keep them.Do not call UI done if:
When a design reference exists, compare against it directly. When no reference exists, compare against the existing product style and nearby components.
All screenshots and browser artifacts must be written under browser/.
Use descriptive names:
browser/<feature>-desktop-before.pngbrowser/<feature>-desktop-after.pngbrowser/<feature>-mobile.pngbrowser/<feature>-error-state.pngbrowser/<feature>-trace.zipDo not commit these files. They are local evidence. Remove them after the task unless the user explicitly asks to keep them.
For implementation work, report in Russian:
Browser checks: flows and viewports tested.Visual evidence: screenshot filenames created under browser/ and whether they were deleted or kept.Functional evidence: interactions/assertions verified.Runtime evidence: console/network/storage/performance status if checked.Fixes made: browser-driven corrections applied.Residual risks: flows not tested, unavailable credentials, external dependency, or manual checks still needed.If browser validation cannot run, state the blocker and the exact fallback used.
browser/ локально.browser-debug.Creates bite-sized, testable implementation plans from specs or requirements, with file structure and task decomposition. Activates before coding multi-step tasks.
npx claudepluginhub nddev-it-com/rldyour-claudecode --plugin rldyour-browser