The Agent Skills Directory

[SAFE]: No malicious code, credential harvesting, or unauthorized network operations were detected in the skill instructions.
[PROMPT_INJECTION]: The skill uses directive language such as "This skill is ALWAYS active" and "Applies to every session, every prompt" to enforce its behavioral rules. While these are instructions to override or prioritize specific agent behaviors, they are designed as guardrails for reliability and do not attempt to bypass safety filters or ethical protocols.
[DATA_EXPOSURE_AND_EXFILTRATION]: The mention of "API key rotation" occurs within the context of Rule 4 (Verification Before Execution), which mandates user confirmation before performing sensitive actions. This is a defensive security practice rather than an attempt to expose or steal credentials.

karpathy-guard