Promptfoo Guardrails is an adaptive AI security system designed to protect LLM-based applications from adversarial inputs and emerging threats. Unlike static rule-based guardrail solutions, it uses a feedback loop between red teaming findings and active defenses to continuously refine its protections over time. Each attempted breach contributes data that strengthens the guardrail system. Key capabilities: - Adaptive guardrails that learn from red team findings and evolve defenses against new and emerging threats - Third-party guardrail validation: can test and validate guardrail systems already in place, acting as an independent verification layer - Deployable in minutes with minimal code changes - Supports cloud and on-premises deployment models - Compatible with all major LLM providers and custom models The product is positioned as both a standalone guardrail solution and a validation layer for organizations that already use other guardrail vendors. Its core differentiator is the continuous improvement cycle driven by actual attack data rather than manually authored static rules. Note: Promptfoo has announced it is joining OpenAI, which may affect the product's future availability and roadmap.