The Agent Skills Directory

[PROMPT_INJECTION]: The skill documentation contains strings commonly associated with prompt injection, such as "Ignore previous instructions" and "Pretend you are DAN". These strings are explicitly placed within the context of a RailsConfig definition as examples of patterns that the guardrail framework should detect and block. They do not represent an attempt to override the agent's instructions.
[EXTERNAL_DOWNLOADS]: The skill references the nemoguardrails package via pip install and provides links to NVIDIA's official GitHub repository. These are well-known and trusted resources for the stated purpose of the skill.

nemo-guardrails