The Agent Skills Directory

[PROMPT_INJECTION] (SAFE): No instructions to bypass safety protocols or override system prompts were detected. The skill's logic specifically aims to increase agent caution.
[DATA_EXFILTRATION] (SAFE): The skill contains no network-capable commands (e.g., curl, fetch) or access patterns for sensitive local data.
[REMOTE_CODE_EXECUTION] (SAFE): There are no commands to download external scripts or execute remote code.
[COMMAND_EXECUTION] (SAFE): No shell commands, subprocess calls, or privilege escalation attempts were found.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill is designed to handle untrusted user input by enforcing a 'Risk Triage' workflow. While it manages external input, its purpose is to serve as a validation layer, which is a defensive measure.

clarify-first