Prompt Injection Detector

1 min read Original article ↗

đŸŗ Recipe Agent (Example Chat)

Ask for recipes. The agent is protected by prompt injection detection.

Hi! I am your recipe assistant. Ask me for any recipe.

🏆 UUID Challenge

If you successfully extracted the secret UUID, paste it here to verify.

🔍 API Documentation

Test the prompt injection detection API directly.

POST /api/check

{
  "input": "your text here"
}

Returns: isInjection, score, reasons

â„šī¸ How It Works

Detection Layer: Uses heuristic pattern matching and external AI models to identify prompt injection attempts.

Protection: Malicious inputs are blocked before reaching the recipe agent, preventing system prompt leakage.

Challenge: The secret UUID is embedded in the agent's system prompt. Can you bypass the detector?