بنقرة واحدة
ui-evaluation
Evaluate the running web app with Playwright E2E and agent-browser visual verification.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
القائمة
Evaluate the running web app with Playwright E2E and agent-browser visual verification.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Build AI chat interfaces using ai-elements components — conversations, messages, tool displays, prompt inputs, and more. Use when the user wants to build a chatbot, AI assistant UI, or any AI-powered chat interface.
Repository development entrypoint for requests like 実装して, 追加して, 修正して, この repo で作って. Handles work artifacts, TDD, smoke/E2E verification, review, and handoff.
Fix GitHub or local review findings, re-verify, and loop up to 3 times before escalating.
Create a PR only after quality gates, work artifacts, and Codex review request preparation are complete.
Repository-specific guardrails for architecture boundaries, work artifacts, durable docs, and review readiness. Use from project subagents.
Repository review entrypoint for requests like レビューして, 監査して, 問題点を見て, 改善点を洗い出して. Runs parallel code, performance/security, and spec-consistency review, then consolidates actionable findings.
| name | ui-evaluation |
| description | Evaluate the running web app with Playwright E2E and agent-browser visual verification. |
| disable-model-invocation | true |
This skill provides the full browser verification flow. CI passing alone is NOT sufficient. Actual browser rendering must be confirmed.
pnpm test:smoke
If E2E fails, STOP. Fix the issue before proceeding.
Requires: agent-browser CLI installed (which agent-browser).
If not installed, skip Phase 2 and note it in the report.
For each route in apps/web/app/routes.ts:
# 1. Open page
agent-browser open "http://127.0.0.1:3000{route}"
agent-browser wait --load networkidle
# 2. Console error check (CRITICAL)
agent-browser errors
# If errors exist → record as FAIL for this route
# 3. Screenshot with annotations
mkdir -p work/qa-reports/screenshots
agent-browser screenshot --annotate "work/qa-reports/screenshots/{route-slug}.png"
# 4. Accessibility snapshot for content verification
agent-browser snapshot
# Verify expected headings/content exist in the snapshot output
For each route verified in Phase 2, additionally evaluate:
Responsive screenshots: Resize viewport and capture at 3 breakpoints:
{route-slug}-mobile.png (375px){route-slug}-tablet.png (768px){route-slug}-desktop.png (1280px)Design checks:
Naming convention: All screenshots follow {route-slug}-{viewport}.png format.
Include a "Design Quality" section in the QA report with PASS/WARN/FAIL per check. WARN items do not block completion but should be noted for follow-up.
Write to work/qa-reports/active/{slug}-browser.md:
work/qa-reports/active/ before completion{route-slug}-{viewport}.png convention