The Agent Skills Directory

[SAFE]: The skill provides structural guidance and documentation templates for developing AI agent skills within the Argus repository. It does not contain executable code beyond benign configuration and structural examples.
[COMMAND_EXECUTION]: The workflow file workflows/create-a-skill.md instructs the user or agent to run uv run ruff check in Phase 5 to validate Python code examples. Ruff is a standard, well-known, and trusted linting tool for Python development.
[SAFE]: The skill documentation correctly identifies and avoids common security pitfalls by promoting the use of explicit boundaries and structured data (though listed as structural advice rather than specific security controls). No patterns of data exfiltration, persistence, or privilege escalation were found.

designing-skills