The Agent Skills Directory

PROMPT_INJECTION (SAFE): Instructional templates in the reference files use standard formatting and do not contain bypass or override patterns.
COMMAND_EXECUTION (SAFE): The Python scripts perform file system operations (reading, writing ZIP files) for their intended purpose without executing arbitrary shell commands.
DATA_EXFILTRATION (SAFE): No network operations or hardcoded credentials were found. File access is limited to the skill directory being processed.
REMOTE_CODE_EXECUTION (SAFE): No external downloads or remote script executions are present. Dependencies are limited to standard libraries and PyYAML.
DYNAMIC_EXECUTION (SAFE): The skill uses yaml.safe_load() to parse configuration, preventing unsafe deserialization attacks. No eval() or exec() calls are used.

skill-creator