The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface as it processes external topics from a pipeline and user-supplied concept definitions. This potential vulnerability is addressed through a mandatory 'Phase 1' logic-locking mechanism that requires the AI to verify theoretical frameworks and definitions against local reference standards before generating content. Additionally, 'Phase 4' mandates a 'Red Line' audit to filter for unauthorized or sensitive terms.
[SAFE]: The skill demonstrates safe behavior by relying on local reference files for its operational logic. All external links are to the author's own documentation on GitHub, which serves a purely instructional purpose. No signs of obfuscation, remote code execution, unauthorized privilege escalation, or data exfiltration were detected.

aikey-concept