The Agent Skills Directory

[SAFE]: The skill uses subprocess.run across several scripts (e.g., at.py, kf.py, plate.py, pool.py) to manage its multi-stage creative workflow. These calls orchestrate the execution of internal runners and media processing tools, which is consistent with the skill's primary purpose as a production pipeline.
[SAFE]: Network operations are directed to reputable and well-known generative AI platforms, including Replicate, Fal.ai, and OpenRouter. These connections are necessary for the skill to perform its intended generative media tasks and are initiated through standard API protocols.
[SAFE]: The skill manages authentication tokens for external services using environment variables (REPLICATE_API_TOKEN, OPENROUTER_API_KEY), adhering to secure development practices for handling sensitive credentials.
[SAFE]: While the skill exhibits an architectural surface for indirect prompt injection—by utilizing vision-model outputs to construct prompts for video generation in motion.py—this is an intended functional requirement for visual continuity in AI video production and is considered low risk in this context.

music-video