The Agent Skills Directory

[SAFE]: The skill demonstrates proactive security measures by explicitly instructing the agent and user to avoid hardcoding API keys, session tokens, or private metadata within project files, recommending the use of environment variables and secure credential managers instead (refer to references/tool-adapters.md and references/continuity-assets.md).
[SAFE]: No obfuscated content, malicious redirects, or hidden commands (such as !command syntax) were detected across the analyzed files.
[SAFE]: The workflow implements a robust 'human-in-the-loop' governance model through mandatory 'Acceptance Gates' and 'Authorized Approver' roles, ensuring that the AI agent cannot autonomously commit to canon without human verification (refer to references/showrunning.md and references/continuity-assets.md).
[SAFE]: The instructions include ethical and safety constraints, such as prohibiting the unauthorized imitation of public figures and requiring compliance with the safety policies of external AI providers (refer to references/casting-character-design.md and references/screenwriting-core.md).
[SAFE]: Data processing is handled through rigid, structured templates which minimize the risk of prompt injection by separating creative instructions from operational data.

ai-short-drama-studio-female