AI agent'lar uchun 20 ta open-source security tool. LinkedIn'da shunaqa ro'yxat chiqibdi 👇
— Prompt injection / guardrails: LLM Guard, NeMo Guardrails
— Red teaming: PyRIT, garak, Promptfoo
— Permission'lar: OPA, OpenFGA, Cedar
— Sandboxing: E2B, gVisor, Firecracker
— Secret va data himoyasi: Presidio, Trivy, Semgrep
Agent web'ni o'qisa, tool chaqirsa, kod ishlatsa, xavf faqat prompt injection bilan tugamaydi. Permission'lar, secret'lar, data ham xavf ostida.
SI'lik app yaratayotganda shularni inobatga olasizlarmi?
Manba: https://lnkd.in/p/dmETjkaj
Post #1737
192
