The Agent Skills Directory

[SAFE]: The skill defines a robust security contract for Bash scripting, mandating the use of set -uo pipefail, quoting variables, and strictly forbidding eval on untrusted input.
[SAFE]: It implements safe file handling patterns, including atomic_write (using temporary files and mv for atomicity) and backup_file (creating timestamped backups before modification).
[SAFE]: The framework includes mandatory validation gates such as require_root, require_family, and require_flag, ensuring that scripts do not execute in unintended environments or with missing parameters.
[SAFE]: It establishes a non-interactive contract for AI agents (using --yes) that requires explicit decision flags, preventing silent or accidental defaults for destructive operations.
[SAFE]: The skill provides cross-distribution support by abstracting package and service management into primitives, reducing the risk of hardcoded commands failing on different Linux families.

linux-bash-scripting