bedrock-guardrails

SKILL.md

Amazon Bedrock Guardrails provides six safeguard policies for securing and controlling generative AI applications. It works with any foundation model (Bedrock, OpenAI, Google Gemini, self-hosted) through the ApplyGuardrail API, enabling consistent safety policies across your entire AI infrastructure.

Automated Reasoning requires a separate policy document, then association with guardrail.

| NONE | No filtering | When policy doesn't apply to direction (e.g., output for PROMPTATTACK) | | LOW | Lenient filtering | Creative applications, minimal restrictions | | MEDIUM | Balanced filtering | General-purpose applications | | HIGH | Strict filtering | Enterprise, compliance-critical applications |

全面的 Amazon Bedrock Guardrails 實施可實現 AI 安全，具有 6 項保障策略（內容過濾器、PII 編輯、主題拒絕、單詞過濾器、上下文基礎、自動推理）。在實施內容審核、檢測即時攻擊、防止幻覺、保護敏感數據、執行合規性策略或通過數學驗證保護生成式 AI 應用程序時使用。 來源：adaptationio/skrillz。

查看原文