The Agent Skills Directory

[SAFE]: The skill uses local Python scripts to automate project management tasks such as initialization, validation, and exporting. These scripts, including project_tool.py and creator_markdown_check.py, are bundled with the skill and operate strictly on local project files within a defined workspace.
[COMMAND_EXECUTION]: The skill executes internal utility scripts to validate Markdown structure and manage project metadata. These operations are essential to the skill's workflow for coordinating multiple documents (scripts, storyboards, prompts) and do not involve untrusted remote code or unauthorized privilege escalation.
[DATA_EXFILTRATION]: The project_tool.py script includes an export command that allows the user to copy project artifacts to a specified external directory for final delivery. This is a legitimate functional feature and includes safeguards to ensure the destination path is safe and does not conflict with the project root.
[INDIRECT_PROMPT_INJECTION]: The skill processes creative content from Markdown files through validation scripts. While this content is user-controlled, the scripts use safe parsing methods such as regular expressions and JSON parsing to verify structure and consistency, and do not pass raw input into sensitive sinks that would lead to safety guideline bypasses.

short-drama