The Agent Skills Directory

[Architectural Security Guards]: The skill describes a series of automated checks designed to enforce architectural boundaries and prevent regressions. This includes specific tests for database boundaries, container workspace isolation, and platform-specific tool locking to ensure consistent environments.- [Command Execution Logic]: The documentation references local audit scripts, such as lint_harness.py and check-surfaces.py, which are used to execute linting tools. These components are described as standard parts of the build and verification system, intended to run in a 'fast phase' to provide early feedback.- [Adversarial Pattern Testing]: The text discusses testing the guard system against adversarial inputs to ensure robustness. It specifically addresses shell metacharacters and control operators (e.g., ;, &, |) that could be used to bypass exit status checks, demonstrating a focus on hardening the enforcement mechanisms.

citadel