The Agent Skills Directory

[COMMAND_EXECUTION]: Defines a safe fallback ladder for privilege elevation, prioritizing sudo and gsudo while requiring manual user intervention as the final resort. It explicitly warns against using methods that would cause the agent to lose control over process output or exit codes.
[SAFE]: Includes defensive procedures for destructive file operations, mandating absolute path resolution and verification before deletion to prevent accidental data loss.
[SAFE]: Provides responsible guidelines for browser automation during identity flows, instructing the agent to hand off sensitive authentication steps to the user rather than attempting to bypass anti-automation or security filters.
[SAFE]: Recommends standard practices for non-interactive Git operations and troubleshooting third-party tool failures using official release information.

command-execution