OWASP Top 10 for LLM Applications

What it is

If your product calls an AI model, it has a new class of risk. This list covers the most important ones.

We look at how prompts are built, what data is sent, and whether the AI's answers are trusted without checks.

Who needs it

Products with chatbots, AI features or agents.

What Good Vibes checks

Prompt injection, data leaking to AI models, trusting AI output. Every finding names the file and line, and comes with a Fix Pack prompt for your AI tool.