The Agent Skills Directory

[COMMAND_EXECUTION]: The provided script scripts/security_check.py uses the Python subprocess.run module to execute security scanners (gitleaks, trivy, semgrep, etc.) and git commands. These calls use argument lists rather than shell strings, which is a standard and secure implementation for this functionality.
[EXTERNAL_DOWNLOADS]: The skill guides users to install necessary security tools through official and well-known package managers such as Homebrew (brew), Python Package Index (pip), and Windows Package Manager (winget). No direct downloads of unverified scripts or execution of remote payloads were detected.
[SAFE]: The skill implements several robust security safeguards:
Interactive Bypass: The --force bypass mechanism in scripts/security_check.py requires a TTY (interactive terminal) and an explicit string input ('YES'), preventing automated or silent bypasses in CI or automated environments.
Destructive Action Guards: The skill's instructions require the agent to ask for user confirmation and perform diffs before overwriting critical configuration files like .pre-commit-config.yaml.
Offline-First Priority: The architecture prioritizes local scanners and warmed databases over cloud-based services, minimizing data exposure and network dependencies.

security-setup