AgentShield - 36/36 tests passed. One-line integration. Protects OpenAI, Anthropic, LangChain against 19 attack vectors including prompt injection, homoglyph, split token, Base64 encoding, and RAG poisoning. Name cannot be more than 100 characters
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).