3 open-source checks before pushing AI-generated apps to production (Security, Browser QA & UI DNA)

Stop pushing AI-generated apps to production without running these 3 open-source checks first:

🛡️ vibe-shield — 92-rule static security scanner & MCP guard (catches leaked Supabase service_role keys in client code, unauthenticated Next.js API routes, IDOR bugs, prompt injection sinks & SKILL.md poisoning in ~1s):
https://github.com/avivgerz-hue/vibe-shield

👻 ghost-qa — Autonomous headless Playwright browser QA (tests your live app across 8 traps: console errors, failed network calls, mobile layout collisions, <44px touch targets, WCAG accessibility & dark/light mode parity):
https://github.com/avivgerz-hue/ghost-qa

🎨 unslop — 72-rule UI Design DNA linter & token exporter (strips generic AI purple gradients & bad typography, enforces 10 studio design archetypes, and exports Tailwind v4/v3, CSS & W3C tokens):
https://github.com/avivgerz-hue/unslop

All 3 are 100% free, MIT-licensed, live on the Official MCP Registry (io.github.avivgerz-hue/*), and work with Claude Code, Cursor, Windsurf, OpenCode & VS Code.

⭐ Star the repos on GitHub to save them for your next build!