The Agent Skills Directory

[SAFE]: The skill contains purely instructional content focused on best practices for cross-platform stability. It guides the agent to perform environment discovery rather than executing unverified third-party code or performing network exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing data from the local environment by instructing the agent to execute system commands and parse their outputs (e.g., uname, lockfile existence, tool versions). While this is the intended purpose, it represents a potential ingestion point for data from a compromised environment. * Ingestion points: System command outputs (e.g., uname -s, systeminfo, lockfile checks, runtime version strings) and environment variables (e.g., $SHELL, $VIRTUAL_ENV). * Boundary markers: None specified in the instructions for isolating detection output. * Capability inventory: Extensive shell command mapping provided for Linux, macOS, CMD, PowerShell, and Git Bash, including file operations (ls, rm, find), process management (ps, tasklist), and network probes (lsof, netstat). * Sanitization: The skill does not explicitly describe sanitization methods for command output before processing.

environment-awareness