The Agent Skills Directory

[SAFE]: The skill functions as a security validation middleware for AI agents, providing a robust framework for input and output sanitization. Its operations are local and focused on protection rather than exploitation.- [PROMPT_INJECTION]: Static analysis flags instruction override and jailbreak patterns in the documentation and API reference. However, these strings are documented as example cases for the validation engine to detect and block, and do not represent a threat to the execution environment.- [EXTERNAL_DOWNLOADS]: The skill requires standard, well-known libraries such as NVIDIA NeMo Guardrails, Guardrails AI, and Microsoft Presidio. These are reputable open-source tools commonly used for AI safety and data privacy.- [DATA_EXFILTRATION]: While the skill contains regex patterns for identifying sensitive credentials like AWS keys and SSNs, these are used strictly for local redaction and sanitization purposes. The provided scripts do not contain network functionality or mechanisms to exfiltrate data.

implementing-llm-guardrails-for-security