AI writes bugs.
We catch them.

AINAScan detects 33 security vulnerabilities + 15 vibe-coding patterns in AI-generated code — deterministic AST, no LLM, always the same result.

9 languages 48 patterns 33 security + 15 vibe-coding No code storage GitHub Action ready

🎉 Promo (until June 24): Use key vg_free_test for full Pro access — no signup required.

# Try it in 30 seconds
curl -X POST https://pleasing-transformation-production-90c2.up.railway.app/v1/scan \
  -H "X-API-Key: vg_free_test" \
  -F "file=@your_file.py"

What vibe-coding bugs look like

MISSING_WRITE

Save function with no write

AI generates save_user(data) that validates input but never runs INSERT or UPDATE.

FAKE_ASYNC

async with no await

async def fetch(): that never awaits — runs synchronously, blocking the event loop.

STUB_SKELETON

Placeholder functions

Functions that just return {} or return None with no real logic.

DEAD_CALL_RESULT

Ignored return values

Calls 3 external services, discards all results, returns a constant.

HARDCODED_TABLE

Dict instead of DB query

40-key hardcoded dictionary where a DB query belongs.

+ 10 more

SQL injection, SSRF, path traversal…

Plus 33 standard security patterns across 9 languages.

How it compares

	AINAScan	Bandit	Semgrep	CodeRabbit
Vibe-coding patterns (15)	✅ Dedicated	❌	❌	❌
Deterministic AST	✅	✅	✅	❌ LLM
No code storage	✅	✅	✅	❌
9 languages	✅	❌ Python only	✅	✅
GitHub Action	✅	✅	✅	✅
Free tier	✅ 50 files/day	✅	limited	❌

GitHub Action — PR security gate

# .github/workflows/vibeguard.yml
name: VibeGuard Security Scan
on: [pull_request]
jobs:
  scan:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Moonsehwan/aina-vibeguard-action@v1
        with:
          api-key: \${{ secrets.VIBEGUARD_KEY }}
          fail-on-block: 'true'

View on GitHub API Status GitHub Action

AI writes bugs.We catch them.