Home
AgentGuard is an embeddable, enterprise-grade AI security framework designed for Java 21 and Spring Boot 3.3+. It acts as a bidirectional security firewall between your application services and Large Language Model providers (OpenAI, Anthropic Claude, Google Gemini, Ollama, Spring AI, and LangChain4j).
As enterprises accelerate Generative AI adoption, production systems encounter three critical threats:
- Confidentiality & Compliance: Unintentional leakage of Personally Identifiable Information (PII), customer credentials, and payment card numbers across third-party API boundaries.
- Integrity & Prompt Security: Injection attacks overriding system guardrails, jailbreaking personas, and synthesizing malicious tool actions.
- Availability & Resource Control: Infinite tool-calling recursion, unbudgeted token consumption, and denial-of-wallet risks.
AgentGuard resolves these challenges with zero remote dependencies, sub-millisecond in-memory execution, and seamless Spring Boot ergonomics.
