The Agent Skills Directory

[SAFE]: The skill is a security utility intended for auditing other skills. The identified malicious patterns are part of its detection logic and reference material, rather than being part of the skill's operational instructions.
[PROMPT_INJECTION]: The file references/prompt-injection-patterns.md contains numerous jailbreak and instruction override examples. These are used as detection signatures for the scanner and do not attempt to manipulate the current agent session.
[REMOTE_CODE_EXECUTION]: The references/dangerous-code-patterns.md and scripts/scan_skill.py files contain code fragments related to arbitrary execution and exfiltration. These are processed as static text for regex-based auditing and are not executed by the skill.
[COMMAND_EXECUTION]: The skill requires Bash permissions to execute its bundled Python-based scanner using the uv package manager. This use of the shell is legitimate and necessary for the tool's primary function.

skill-scanner