The Agent Skills Directory

[SAFE]: The skill instructs the agent on processing .allium specification files. It contains no executable scripts, packages, or binary dependencies.
[SAFE]: The interaction modes are well-defined for both interactive and non-interactive environments, focusing exclusively on domain modeling and system behavior translation.
[SAFE]: References to local documentation (e.g., ../allium/references/language-reference.md) and standard CLI verification commands (e.g., allium check) are handled safely within the skill's defined boundaries.
[SAFE]: The skill explicitly pushes back on vague requirements and establishes boundaries that prevent editing execution code, restricting modifications solely to .allium spec files.

tend