barretocyber/bonklm---llm-security-...

BonkLM is a comprehensive Node.js security library that protects AI applications from prompt injection, jailbreaks, data leaks, and other malicious attacks. It provides framework-agnostic, provider-agnostic guardrails for LLM applications with multiple validation layers including prompt injection detection, jailbreak prevention, secret and PII guarding, and real-time streaming validation. The project is designed for developers building AI applications who need production-ready security controls to safeguard their LLM integrations from common attack vectors.