The Agent Skills Directory

[SAFE]: The skill implements defensive logic designed to protect sensitive files (such as .env, credentials, and SSH keys) and prevent destructive shell commands (like rm -rf or DROP TABLE). These patterns are identified as security features (Tier 1 reflexes) rather than vulnerabilities.
[SAFE]: References to external repositories and domains (such as github.com/itallstartedwithaidea/agent-skills and googleadsagent.ai) belong to the skill's author and represent standard documentation and project links.
[SAFE]: The provided code snippets demonstrate a structured approach to agent safety, focusing on pre-action validation and intent filtering. No malicious patterns, such as obfuscation, credential exfiltration, or unauthorized command execution, were found.

agent-instinct-system