The Agent Skills Directory

[SAFE]: The skill defines a comprehensive safety framework (GAOS) for managing agent execution, including input sanitization, policy constraints, and execution sandboxing.\n- [SAFE]: It explicitly instructs the agent to block destructive operations (e.g., 'rm -rf', 'git push --force') and unauthorized privilege escalation (e.g., 'sudo', 'chmod 777').\n- [SAFE]: The skill includes robust guidelines for detecting and redacting sensitive information, such as API keys, credit card numbers, and PII, from agent outputs.\n- [INDIRECT_PROMPT_INJECTION]: While the skill establishes a framework for processing untrusted user data in Layer 1, it provides detailed sanitization and boundary enforcement rules that prioritize safety and mitigate potential injection risks.\n- [SAFE]: No executable code, remote script downloads, persistence mechanisms, or obfuscated content were found within the skill file.

governance-gate