The Agent Skills Directory

[SAFE]: The skill is composed entirely of markdown instructions designed to guide an AI agent in generating structured text prompts for video generation. It does not include any scripts, executables, or command-line operations.
[PROMPT_INJECTION]: While the skill uses imperative language such as "HARD LOCK" and "LOCKED ORDER," these are internal constraints to ensure output consistency for the user's task. No evidence of jailbreak attempts, safety filter bypasses, or instructions to ignore previous system prompts was found.
[DATA_EXFILTRATION]: The skill does not contain any network-reaching commands (e.g., curl, wget) or hardcoded credentials. It does not attempt to access sensitive local file paths like SSH keys or environment variables.
[REMOTE_CODE_EXECUTION]: There are no package manager configurations (npm, pip) or patterns indicating the download and execution of remote code.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing user-provided data (images and descriptions). However, because the skill's capabilities are restricted to text generation and it lacks the ability to execute code or perform network operations, this surface does not present a viable attack vector.

cinema-worldbuilder