The Agent Skills Directory

[SAFE]: The skill restricts its environment to non-modifying tools (Read, Grep, Glob), ensuring it cannot execute arbitrary code or modify the host system during its analysis process.
[SAFE]: Contains a dedicated safety notice that instructs the agent to treat target content as data rather than instructions, specifically warning against executing code or following prompts found within reviewed files.
[SAFE]: No obfuscation, hidden URLs, or credential harvesting patterns were detected. All external references point to reputable security frameworks (OWASP, NIST) or public security research repositories.
[SAFE]: The skill methodology aligns with the principle of least privilege, advising users to scope agent permissions and use human-in-the-loop gates.

agent-security