The Agent Skills Directory

[SAFE]: The skill focuses on documentation and UI specification tasks. It uses local templates (assets/admin-screen-spec.template.md) and reference contracts (references/admin-ui-build-contract.md) to guide the creation of admin screens.\n- [COMMAND_EXECUTION]: The skill includes validation steps in SKILL.md that execute local Python scripts (quick_validate.py, lint_production_skill.py) located within the environment's $CODEX_HOME directory. These are standard development tools used for linting and validating the skill's structure.\n- [DATA_EXFILTRATION]: There is no evidence of hardcoded credentials, sensitive file access outside the skill's scope, or unauthorized network requests to external domains. The safety rules explicitly instruct the agent to avoid exposing secrets or private data.\n- [PROMPT_INJECTION]: The instructions do not contain any patterns typical of jailbreaking or overriding safety filters. The use of clear operating modes and workflow steps maintains the agent's focus on its intended purpose.

admin-ui-builder