The Agent Skills Directory

[SAFE]: The skill functions as a coordinator, providing instructions for the sequential execution of other locally stored skills using the local filesystem as a communication channel.
[COMMAND_EXECUTION]: Utilizes ffmpeg for video slicing and post-processing, and bun/remotion for programmatic video rendering. These tools are explicitly restricted in the allowed-tools section and are used for their intended media production purposes.
[SAFE]: Implements strict 'Human-in-the-Loop' gates between every phase of production (Script, Design, and Video), ensuring the agent cannot auto-advance or perform actions without explicit user approval.
[DATA_EXPOSURE]: Accesses project-specific files in the brand/ and marketing/ directories and internal skill configurations in ~/.claude/skills/. This localized file access is necessary for the content pipeline and does not involve exfiltration to external domains.

tiktok-slideshow