The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a mandatory protocol for agents to run Python-based verification scripts located within the local .agent/skills/ directory (e.g., ux_audit.py, api_validator.py, security_scan.py). These commands are part of an internal quality control framework used to validate code changes before completion.
[PROMPT_INJECTION]: The instructions include directive language such as 'CRITICAL SKILL', 'RED RULE', and 'MANDATORY' to enforce coding style and verification steps. These are standard operational instructions and do not contain patterns intended to bypass safety filters or extract system prompts.

clean-code