The Agent Skills Directory

[SAFE]: The skill provides a structured framework for quality control and does not contain any malicious code, obfuscation, or unauthorized access patterns. It includes security-positive recommendations such as including security and injection prevention in the evaluation rubrics.- [COMMAND_EXECUTION]: The skill describes the use of platform-specific subagent capabilities for content verification. These operations are confined to the agent's internal orchestration layer for generation and review rather than arbitrary system command execution.- [PROMPT_INJECTION]: The framework ingests untrusted data (the output under review) into subsequent agent prompts. While this presents an indirect prompt injection surface, the framework uses structured evaluation rubrics and explicit instructions for agents to be adversarial and identify problems, which serves as an effective mitigation.

santa-method