The Agent Skills Directory

[SAFE]: No malicious patterns, suspicious obfuscation, or safety bypasses were detected in the skill instructions. The skill operates within a controlled framework for content auditing.- [PROMPT_INJECTION]: The skill is designed to process untrusted user-provided video briefs. While this creates a surface for indirect prompt injection, the analysis confirms the following: (1) Ingestion points are limited to the user's content plan. (2) No specific boundary markers are present to delimit user input. (3) The skill's capabilities are restricted to internal routing to video generation tools (video-gen, video-engine-routing). (4) No sanitization is performed on the input. This is a low-risk, standard implementation for this type of utility.- [REMOTE_CODE_EXECUTION]: The skill references internal routing targets and specific video engines like Higgsfield for identity consistency, but it does not perform any remote script execution or external package installations.

production-review