一键导入
design-review
Проведи дизайн/visual review через управляемый CloakBrowser boundary. EN: review design implementation with managed browser evidence.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Проведи дизайн/visual review через управляемый CloakBrowser boundary. EN: review design implementation with managed browser evidence.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Проверь browser/UI поведение только через управляемые Playwright CLI и Chrome DevTools MCP. EN: validate browser behavior through CloakBrowser.
Воркер cmux v3 для Antigravity CLI: cmux worker role, typed task envelope, heartbeat, scoped JSON report, no push/final sync. Используй для: cmux worker, worker report. EN triggers: cmux v3 worker. EN: cmux v3 worker.
Инициализируй Antigravity CLI adapter scope без изменений. EN: initialize Antigravity CLI adapter scope read-only.
Проведи read-only Antigravity rldyour review. EN: read-only Antigravity rldyour review.
Выполни полный Antigravity rldyour task lifecycle. EN: full Antigravity rldyour task lifecycle.
Синхронизируй Antigravity docs, memories и git. EN: synchronize Antigravity docs, memories, and git.
| name | design-review |
| description | Проведи дизайн/visual review через управляемый CloakBrowser boundary. EN: review design implementation with managed browser evidence. |
Evaluate visual implementation against Figma, screenshots, reference images, or written specifications.
This boundary applies before every browser action:
Run exactly:
$HOME/.local/bin/cloakbrowser-cdp-health
If the command is missing or exits nonzero, stop immediately and report NOT_PROVEN.
Browser execution is permitted only through:
$HOME/.local/bin/playwright-cli executable; run-code and --filename are forbidden;/bin/sh -c 'exec "$HOME/.local/bin/chrome-devtools-mcp" --headless --isolated --no-usage-statistics --no-performance-crux'.Never execute the Webwright Python runtime, stock/raw/in-app Browser, browser_agent, node_repl, computer-use, Playwright MCP, raw Playwright, bunx, npx, direct package invocations, alternate CDP endpoints, alternate browser executables, alternate browser configs, or any fallback. No fallback is allowed.
webwright-task is compatibility intent routed to browser:validate; it never
authorizes Webwright runtime execution.
Use for Figma-to-code review, responsive visual checks, typography, color/token, layout, and accessibility deviations.
Expected source, target URL, viewport matrix, dynamic masks, and thresholds.
Use Figma MCP for the expected source, managed Playwright CLI for actual screenshots and responsive evidence, and managed Chrome DevTools MCP for DOM/runtime diagnosis. Rerun health immediately before each browser action.
Report expected source, exact artifact paths, masks, thresholds, pixel/layout,
typography/token/accessibility deviations, confidence, and NOT_PROVEN gaps.
Do not approve visual quality without artifacts or bypass the browser boundary.
Run python3 scripts/validate_gemini_browser_routing.py --strict.
Report each mismatch with source, viewport, severity, evidence, and any
NOT_PROVEN state.