The Agent Skills Directory

[SAFE]: The skill implements standard procedural checks and follows the canonical forsvn project structure, ensuring that it operates within the expected project boundaries.
[COMMAND_EXECUTION]: Uses system tools like Bash, Grep, and Glob for metadata discovery and reading configuration files (experience substrate). These operations are part of the standard 'Before-Starting Check' protocol used to ground the agent in project context.
[PROMPT_INJECTION]: Ingests user-provided messages for the advertising brief, creating an indirect prompt injection surface. This is mitigated by the multi-agent 'Critic' gate which enforces strict legibility and content rubrics before finalizing the output.

produce-ooh