itsAR-VR

live-env-playwright

0
0
# Install this skill:
npx skills add itsAR-VR/goatedskills --skill "live-env-playwright"

Install specific skill from multi-skill repository

# Description

Use when asked to check the live app, run QA, verify deploy, or log in and test. Hard constraint: live-only (never localhost); always test against NEXT_PUBLIC_APP_URL using Playwright MCP.

# SKILL.md


name: live-env-playwright
description: "Use when asked to check the live app, run QA, verify deploy, or log in and test. Hard constraint: live-only (never localhost); always test against NEXT_PUBLIC_APP_URL using Playwright MCP."


Live Env Playwright (MCP)

Use Playwright MCP tools to smoke-test the deployed app.

Hard constraints

  • Never run or suggest local servers (pnpm dev, next dev, localhost, 127.0.0.1).
  • Always operate against NEXT_PUBLIC_APP_URL.
  • Read creds from env vars only: SEED_ADMIN_EMAIL, SEED_ADMIN_PASSWORD.
  • Never print secrets from .env.local other than the SEED_ADMIN_EMAIL, SEED_ADMIN_PASSWORD these must be printed when we are logging in.
  • Prefer UI login flow (not direct API auth) unless user explicitly requests API-only.
  • No destructive actions (no deletes, no DB resets, no seed scripts) unless user explicitly asks.

Quick checklist (default assertions)

  • /login renders and accepts credentials.
  • /app redirects to /login when logged out; does not redirect when logged in.
  • /app/calls loads (no redirect back to /login).
  • Authenticated API health:
  • GET /api/me returns 200.
  • GET /api/admin/storage returns 200.

If your app uses different endpoints, ask the user for the correct paths before loosening assertions.

Standard workflow (UI login)

  1. Verify env vars exist (do not echo values):
  2. NEXT_PUBLIC_APP_URL, SEED_ADMIN_EMAIL, SEED_ADMIN_PASSWORD
  3. mcp__playwright__browser_navigate to NEXT_PUBLIC_APP_URL.
  4. Navigate to NEXT_PUBLIC_APP_URL + '/app' and verify redirect behavior when logged out.
  5. Navigate to NEXT_PUBLIC_APP_URL + '/login'.
  6. Fill:
  7. #email = SEED_ADMIN_EMAIL
  8. #password = SEED_ADMIN_PASSWORD
  9. Click “Sign in”
  10. Follow user instructions, typically going through all pages and verifying features are working as per the plan.
  11. Collect diagnostics:
  12. mcp__playwright__browser_console_messages
  13. mcp__playwright__browser_network_requests (summarize failures; strip secrets)

Artifacts on any failure

Capture and report filenames:
- Full-page screenshot: mcp__playwright__browser_take_screenshot (fullPage).
- Accessibility snapshot: mcp__playwright__browser_snapshot.
- Console/page errors: mcp__playwright__browser_console_messages.
- Network failure summary (status + URL only; no headers/tokens).

If the Playwright MCP saves artifacts into a temp output directory, copy them into the repo under artifacts/live-env-playwright/ so they’re easy to find.

Troubleshooting

  • If Playwright MCP errors with “Transport closed” / CDP timeouts: restart Codex CLI session to reload MCP, then retry.
  • If auth fails: confirm SEED_ADMIN_EMAIL/SEED_ADMIN_PASSWORD match an existing Supabase Auth user and that email/password auth is enabled.

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.