The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to ingest and act upon untrusted data from user briefs and the output of media analysis tools.
Ingestion points: Untrusted content enters the context through user-provided project briefs and the results of the $media-analysis tool (referenced in SKILL.md under 'Workflow' and 'Review Mode').
Boundary markers: The skill lacks explicit prompt delimiters or 'ignore' instructions for external content; however, it enforces a highly structured output format (shot tables and specific UGC Direction fields) which acts as a structural boundary.
Capability inventory: The skill has the capability to trigger media generation tools, specifically $image-batch-runner and $video-batch-runner, based on the processed instructions.
Sanitization: No explicit sanitization or filtering of external content is performed; however, the 'Quality Bar' and 'Non-Imitation Boundary' in SKILL.md provide significant safety constraints that prohibit the agent from inventing claims or mimicking specific identities, mitigating the risks of adversarial manipulation.

ugc-ad