The Agent Skills Directory

[SAFE]: The skill defines a governance framework for AI agent tasks. It includes a self-validation script (scripts/validate.sh) that performs local integrity checks using standard tools like grep. No sensitive data access, remote code execution, or obfuscation was found.
[INDIRECT_PROMPT_INJECTION]: The skill processes outputs from planning and implementation phases. It implements guardrails by requiring fresh context for validation and distinct author/validator roles to prevent instruction leakage or bias.

rpi