The Agent Skills Directory

[SAFE]: The skill provides legitimate security patterns and implementation snippets for filtering and tool governance.
[PROMPT_INJECTION]: Static detections of injection patterns like 'ignore your instructions' are false positives, as these strings are used strictly within a defensive context to demonstrate how guardrails should identify and block malicious inputs.
[EXTERNAL_DOWNLOADS]: The skill suggests using established security frameworks like LLM Guard and Microsoft Presidio; no unverified or suspicious external dependencies are present.

agent-guardrails