Red-Team an AI Agent With Realistic Attack Paths

实施步骤

1. Map assets, actors, trust boundaries, and high-impact actions.
2. Create attacks for injection, exfiltration, unauthorized writes, and denial of service.
3. Run tests in an isolated environment with kill switches.
4. Record root cause and add a durable control plus regression case.

验证清单

• Verify controls at model, application, and tool layers.
• Test chained attacks rather than isolated strings.
• Retest after model and tool changes.

常见错误

• Testing only jailbreak phrases.
• Using real customer data in attack fixtures.
• Fixing one prompt without addressing missing authorization.

实际决策

先交付能够通过上述检查的最小设计。记录测试中使用的模型、提示词、工具、数据集与运行时版本，让后续变化可以被比较而不是被猜测。依赖供应商的限制、模型名称、价格或预览功能之前，请重新检查链接文档；这些细节变化往往快于本文的工程原则。

官方来源

• OpenAI, Safety best practices ↗
• Model Context Protocol, Security best practices ↗