The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses specific local CLI commands in SKILL.md (e.g., calling python image_to_image.py) to manage local image rendering pipelines, character blending operations, and file manipulation. These utility operations follow safe development patterns and execute exclusively over local project structures.
[EXTERNAL_DOWNLOADS]: The skill declares structural references and template lookups targeting locally hosted asset folders and repository documents. No unknown network-based retrieval, dynamic code pulling, or untrusted package integration channels are utilized.
[INDIRECT_PROMPT_INJECTION]: The skill ingests raw textual content provided by users (such as raw stories, training scripts, or multi-scene dialog guides). This input configuration constitutes an indirect input parsing interface. However, the system leverages explicit structure parameters, compartmentalized template blocks, and metadata indices that effectively insulate underlying instructions from malicious override patterns.

story-to-scenes