The Agent Skills Directory

[COMMAND_EXECUTION]: The skill specifies shell commands like rg, find, git, and env for the agent to identify secrets in the filesystem, environment, and version history. These capabilities are intended for security auditing purposes.
[PROMPT_INJECTION]: The skill employs persona-based instructions ("SAFE-CHECK — Agente Supremo de Seguranca") to define its operational context. These instructions focus the agent on security best practices and do not contain directives to bypass safety filters.
[SAFE]: The skill includes explicit security controls, such as forbidding the agent from asking for credentials in plain text and providing redaction commands to ensure sensitive data does not persist in the agent's context or conversation history.

cred-omega